For the record, got mine from a Mecha Gnome while exploring, so I guess any gnome has a chance to drop it.
EDIT: It happened AFTER the Vault weekend.
From my understanding the troops and orbs was not suppose to be stacked . So these rewards are absolute trash. The devs are in isolation and as of now have not addressed anything and probably won’t. All we want is a explanation.
Just got off the stream, Kafka has asked the team to ‘verify’ the epic vault key drop rate.
my question would be are they going to verify the normal drop rate (which is probably working as intended) or are they going to verify the drop rate that should have been occurring during the vault weekend.
The vault key drop rate doesn’t change on vault weekends. Only the gnome rate is increased. More gnomes → more vault keys, but the odds of getting a vault key from a gnome are unchanged.
Edit: I was wrong, the drop rate is different during vault events.
We checked the drop rates for the Vault Key and Epic Drop Keys and they were correct.
I’ve passed on the feedback that the low drop rate was frustrating and it will be looked at for the future.
To be clear, does that mean the game handed out X number of total normal + epic vault keys over the weekend (not including pity keys) and 10% (or even any percentage > 5%) of those were epic vault keys? I’ll take your word for it, just reconfirming the interpretation.
that shouldn’t be the case. there’s at least 2 separate instances that mention it should be +gnomes and +key drop chances
Usually special Vault events only touch the appearance rates and ignore drop rates, but drop rates should be higher, in general
Thanks for passing that along; I know that’s all you can do at the moment. It’s not clear if you simulated drops or looked at the code; hopefully the former.
I hope we can organize some serious data tracking for the next vault weekend, with people precommitting to reporting their results. I remain very skeptical that it’s working as intended - it seems at least ten times too rare - but I’ll wait for more solid data.
Good luck to everyone on the next weekend. I’m pretty burnt out after 400 gnomes, but I’ll give it a go again next time.
When something has happened already, I normally go look at actual logs. 2 numbers to compare. A 5 year old game with no extensive statistical logging would be quite puzzling indeed.
I take it back, I remembered a conversation wrong. It was the opposite. From Kafka, on GoW discord in reference to vault events: “it’s increased chances for Treasure Gnomes to appear and for Vault Keys to drop” (discord link)
And a few messages later: “the drops for vault keys in the vault are increased too”.
Every time I read the zendesk article on vaults, I think it scrambles my brain. Sorry. Thanks for the correction.
True, I just wasn’t sure how easily they could pull the total number of each key dropped across all (or several) players from the logs. That would definitely be hard to argue with.
It’s very possible that the drop rate is fine, and that several unlucky people and guilds gravitated to the forum to complain. The very low global achievement numbers are suspicious, though.
These two quotes cannot both be true. I am well aware that @Saltypatra and @Kafka can only pass on what they’ve been told, but a 10% rate would mean that many more players would have at least one Epic Key by the end of the weekend, and some players would have been lucky and gotten 2 or 3.
The coded drop rates might be correctly entered into the code, but something is happening to make the actual drop rates far lower than promised. I can only ask that the devs continue to investigate rather than claiming that things are working as announced when they clearly aren’t.
I did ask for the interpretation; sometimes people read the words and understand differently. Looking at the actual text again, maybe it means
are rarer … about (10%) of players are expected to see 1 in their lifetime?
Tongue in cheek. But, 10% of what?
The very low global achievement numbers are suspicious, though.
Not necessarily. On Google Play it says “Fewer than 0.1% players have this.” That may look small, but then you look at another one - “This Journey Begins…”, which shows 1% of players have this.
So only 1% of the player base in week 5 have completed week 1. In fact, only 2% of players have done ANY task. Let’s say this 1% player base are the more-than-casual to hardcore population, and only 10% of them actually participate meaningfully in Vault event. Your entire percentage of installed base that can make that achievement is down to 0.1% already. Theoretically, this 0.1% base are able to achieve 120 gnomes > 10 keys > 1 epic. The others are just unlucky.
Of course the code is correct. That’s all they’re willing to bother checking. “Yep, says 10% working as intended” But will they check the total weekend data of all vk’s and evk’s dropped or run some simulations? Of course not that’s too much effort.
It’s the vorpal upgrade all over again which was silently fixed months after them saying it was “working as intended.”
Can you go a little bit more in depth on how they were verified to be correct?
For example: “We simulated 1,000,000 gnome drops and Vault Keys were about 10 times as common as Epic Vault Keys” versus “We looked at all the data collected from a Vault Event weekend and there were approximately 10 times as many total normal vault keys dropped as Epic Vault Keys, as expected” versus “We checked the code” versus “we ran any of the above checks and it was all what was intended, which may or may not be the same as what was communicated?”
Total vault keys obtained versus epic vault keys from gnome drops only over the vault weekend specifically from live data would be the absolute the best indicator possible here, if there is any feasible way to get that data on your end. If simulated data is used, the next best thing would be to also simulate a vault weekend, which may or not have been done. Just “looking at the code” or “verifying we input the right numbers for rate” are historically proven as not being good indicators of verifying something is working correctly. There is also the chance that everything is working exactly as intended, but the number communicated to the playerbase in the patch notes does not align with what is actually in game.
So far, all data points available on the player end are pointing toward something being “off”. In perspective, not quite as “off” per the data that we have as when it was shown that Chaos Portal drops were wrong, but more “off” than when it was shown that average event point multiplier was wrong in the “From The Depths” world event, both of which the actual rates ended up aligning remarkably well with the data collected from the playerbase. Still possible for every data set we have to just be “bad luck” and all of them combined to be an extreme outliers, but nobody wants another situation where the official stance is “it is correct, we checked” but then was some unforseen complication between what was “checked” that allowed the check to be validated but does not explain why the data collected did not even remotely align with expectations.
Thing is, only a small portion of players that installed game, are actively playing (and Steam statistics count all those players for total %). From that part, only a small percentage visits forums and from that group yet another small portion of players are posting on forums.
If anything, a group of 400 active forum members might produce any statisticaly-usefull data, after few weeks/months… not a single vault weekend.
From my point of view - congratulate these who were lucky and got their epic key and keep calm and wait next few weeks to see, if you get any. If you want - you can collect your data on the way and contribute to some player- shared statistics.
if something has a 1% chance to occur, sometimes it takes more than 200 to see it once… and sometimes you can see 3 of such things appear in like 10 tries, and than nothing for another 500 tries…
heck, even if coin has a 50% chance to flip one side or another, you still aren’t assured, it wont flip same side many times in a row… you probably would have to toss a coin more than 50 times, to see these percents near 50, and still it would probably be skewed… one way or another
Who cares? It will happen. Let it roll
My stats professor made a good demonstration about this. He tells us he’s going to flip the top card of from this deck of cards, marking if it is red or black one at a time.
Red, Red, Red, Red, Red, Red, Red, Red, Red, Red, Red, Red, Red, Red…
The class goes insane in astonishment and he asks the question. How much evidence do you need to start doubting that actually am not using a normal deck of cards? As he shows us the rest of the deck which consist of only red cards.
I personally understood that epic vault keys are to be 10% of the drops when a vault key is rolled from a gnome. Did I misunderstand this statement?