PlanetCalypsoForum.com :: Entropia Universe Discussion and Resources
Page 3 of 4 FirstFirst 1 2 3 4 LastLast
Results 21 to 30 of 39
  1. #21
    the thing is that this test has some medium sized faulty factors in it.
    for example it seems that they dropped directly after each other. given the size of the planet and the amount of players id argue, that realistically nobody drops a few seconds after someone else. a more ralistic approach would be to drop like 2 minutes later. best would be to do the same test with 1 minute break, then 2 minutes, then 5 minutes, 10 minutes and then 15 minutes.
    next thing id argue is what happens if the finders have a different search depth and not overlap by 100% like everyone using the same finder. id say the chance that the miner before you used the exact same finder with exact same enhancers is medium to low. so there should be above test done with different finders as well. maybe even with same finders but different enhancers.

    the problem is we dont know how the system determines if you get a hit or not. and as we dont know it yet we need to test out which of the possible factors reduces the hitrate. in this example many of them havent been tested. so the only thing this test shows if if you have the exact same finder as the person before you and you drop just seconds later than him then the hitrate will most likely be pretty bad.

    on top of that 30 or 50 drops is not really enough. e.g. the it could be more like condition crafting where you probably get a lot higher claims and probably have a higher chance for a big payout. but this would have need to be done over thousands of drops, that probably nobody will ever make. one hint that this might be true is that person A gets a 26% hitrate with 33 ped TT while the other miners have about a fifth of that hitrate but only have about half to a third of the TT value while it should be a fifth as well so there is definitely something different. sadly this hasnt been tested out further.

  2. #22
    Alpha kingofaces's Avatar
    Joined
    Jun 2013
    Gender | Ingame
    Male | Male
    Location
    US
    Avatar
    Tony KingofAces Hans
    Posts
    577
    Images
    10
    Quote Originally Posted by GxB View Post
    This quote is hidden because you are ignoring this member. Show Quote
    the thing is that this test has some medium sized faulty factors in it. . .
    Most of these incorrect assumptions have already been addressed in this thread or the followup that we planned to do after this. To reiterate a few though:

    on top of that 30 or 50 drops is not really enough
    This has been an extremely common mistake in these threads, which is why I bring it up first (more details on why this in incorrect is in other replies) You don't directly compare averages with just 10s of whatever sample size you're working with. No one did that here though. Most people know that, but most don't know you don't literally go and do thousands of samples. Most any intro research statistics course goes over these details, but the statistical tests used here for those averages are designed to approximate what that true average would be at say 10,000 samples based on a smaller sample size and determine how likely a legitimate difference is between treatments as opposed to random chance.

    That's why us scientists use them in almost any peer-reviewed articles. 30-50 drops per treatment is already much higher than many published studies that would only use 10-15 as a sample size for a simple experimental design like this one. ~30 is generally considered pretty good with data that has variation like this. Being binomial (hit or miss) helps a lot since those distributions need smaller sample sizes than something closer to 100-300 samples to detect small differences in TT for things like claim size vs. finder decay. There's also not really any evidence that we have power issues since that deals more with false negatives as opposed to false positive. Since we detected a difference, that would instead be a question of false positive or not, and that's built into the tests already where increasing sample size doesn't really change that at this point. That both miners 2 and 3 had lower hit rates just makes the case that a false positive there is even less likely.

    tl;dr statistical significance is what tells the story here, not so much sample size or raw averages even though they play in to those tests.

    for example it seems that they dropped directly after each other.
    I suggest rereading the first posts of this thread. The intent of this was to determine if other miners can affect your hit rate at all. Of course testing was going to occur at the smallest interval where an effect was most likely. People were claiming this didn't happen at all, so the first step was to try to force the effect with a narrow interval. After that, the second round of testing was looking at how long that effect last for, which looks to at least be < 15 minutes. That means the effect generally only is going to be seen "in the wild" at places like Ashi or other small high volume areas (maybe right around TPs).

    next thing id argue is what happens if the finders have a different search depth and not overlap by 100%
    I have some later testing between and F-106 w/ 7 enhancers vs. no enhancers, and claims are practically nonexistent on the second drop if you drop once shallow then drop deeper the second time anyways, so it's not really a depth overlap issue. If depth overlap mattered, the results in the graph in this thread would have looked different than just the initial decrease, and that depth testing I just mentioned more or less replicated the graph in this thread. It basically looks like the z-plane (depth) doesn't affect hit rate in general, but rather the x-y plane. Depth is more about resources type (and generally better MU).

    That didn't matter here though since we were looking at what the rough respawn rate was if you were dealing with the situation in this round 1 thread. At the time, that was before my later testing, so testing at the same depth would have accounted for the same scenario if depth overlap mattered just to be on the safe side. If we had varied depth in this or round 2, that would have been a confounded experiment, and someone could have said the hit rates differences or lack thereof could have just been due to an untested depth effect. This design instead accounted for depth.

    the problem is we dont know how the system determines if you get a hit or not.
    For the purposes of these two rounds of testing, that doesn't matter. All the mattered is what you do in any basic experimental design. Standardize what you can, and account for uncontrolled variability in the rest of your experimental design. We deal with "black boxes" all the time in science, so this was just testing what happened to average hit rate as you changed time between drops. If the baseline hit rate of that area changed over time, that would have been accounted for with everyone mining roughly sequentially (i.e., blocking by time). That also applies to round 2 to a degree.

  3. #23
    did you do the f106 with and without enhancers drops urself or over 2 persons, one with enhancers and one without? otherwise that test is also flawed/ needs to be repeated with the other at least

    and yes, i know what you wanted to show and thats fine. but reading through some replies here it seems that many think that this is now over and they know everything. thing is, this test just shows a tiny part without countertesting different variables. but this needs to be done or the whole hypothesis could fall over.

    i dont mean to discredit your testing, i just wanne show that there is tons of stuff missing to understand this mechanic. i mean sure, one step at a time but when those steps arent taken at all then we cant really say much.
    for example what if the hitrate is completely normal, when dropping just 1m apart from the first miners drop. or what if one minute later, or 5 minutes later the hitrate is completely normal? then your test is nice to see that there is some connection but in reality it wouldnt have any impact at all. and important to test is how it affects the reality of a miner as well

  4. #24
    Alpha kingofaces's Avatar
    Joined
    Jun 2013
    Gender | Ingame
    Male | Male
    Location
    US
    Avatar
    Tony KingofAces Hans
    Posts
    577
    Images
    10
    Quote Originally Posted by GxB View Post
    This quote is hidden because you are ignoring this member. Show Quote
    did you do the f106 with and without enhancers drops urself or over 2 persons, one with enhancers and one without? otherwise that test is also flawed/ needs to be repeated with the other at least
    That was just by myself. There's no indication in either set of data that the results change whether you have multiple people dropping like we did here or if you do it yourself. You still get a statistically significant lower hit rate on the second shortly-after drop regardless, so that's more replication on this that I've been doing while testing other things. It's possible to still get a claim on the second drop, but that's the point between when the radius has been cleared and when whatever the respawn rate is has brought claims back to whatever "full" density is, hence the reduced by non-zero hit rate.

    or what if one minute later, or 5 minutes later the hitrate is completely normal?
    I linked to it earlier, but we tested exactly that in round 2. You generally don't see reduced hit rate in that area after 15 minutes. It's possible the respawn rate varies either by location (respawn of specific resources is a slightly different topic), but that's up to individuals to test at this point.

    Basically, a lot of your questions are things we already accounted for when designing these tests.
    Last edited by kingofaces; 05-23-2020 at 00:18.

  5. #25
    Quote Originally Posted by kingofaces View Post
    This quote is hidden because you are ignoring this member. Show Quote
    That didn't matter here though since we were looking at what the rough respawn rate was if you were dealing with the situation in this round 1 thread.
    so did only the first droper get claims while everyone else who droped in the same location right after him did get nothing at all?

    If no, then i highly doubt that spawns/respawns is a thing...
    the payback hof is a lie

  6. #26
    Alpha kingofaces's Avatar
    Joined
    Jun 2013
    Gender | Ingame
    Male | Male
    Location
    US
    Avatar
    Tony KingofAces Hans
    Posts
    577
    Images
    10
    Quote Originally Posted by Alukat123 View Post
    This quote is hidden because you are ignoring this member. Show Quote
    so did only the first droper get claims while everyone else who droped in the same location right after him did get nothing at all?

    If no, then i highly doubt that spawns/respawns is a thing...
    Just to be clear on which is being discussed, that's in this thread (round 1), and all in the original graph:



    Those are non-zero hit rates, but still significantly reduced even after the first person dropped until the entire radius was empty. Miner 2 was 5 minutes after miner 1, and miner 3 was 10 minutes after miner 1.

    In round 2, we basically repeated this, but extended the wait time between drops. With the second miner being 15 minutes after the first "clearing" miner, there was no difference in hit rate:



    Edit: grabbed the wrong graph before, should have been hit rate, not TT

    Basically, something fairly consistently was happening between 10 and 15 minutes where the respawn rate or whatever you want to call it was adding a few claims early on, but didn't return it to normal hit rate until about 15 minutes without any mining there prior to the last drop. Regardless of what it's called, hit rate drops after a spot is initially mined, and "something" replenishes it over time at whatever rate that is.
    Last edited by kingofaces; 05-23-2020 at 00:59.

  7. #27
    but why is the tt from second and third dropper not a fifth of first dropper, while the hitrate is a fifth for both (approx)? it seems the hitrate is lowered by 75ish % but the tt return just by 50ish%. so theres some correlation there. would be interesting to test this further

  8. #28
    Alpha kingofaces's Avatar
    Joined
    Jun 2013
    Gender | Ingame
    Male | Male
    Location
    US
    Avatar
    Tony KingofAces Hans
    Posts
    577
    Images
    10
    Quote Originally Posted by GxB View Post
    This quote is hidden because you are ignoring this member. Show Quote
    but why is the tt from second and third dropper not a fifth of first dropper, while the hitrate is a fifth for both (approx)? it seems the hitrate is lowered by 75ish % but the tt return just by 50ish%. so theres some correlation there. would be interesting to test this further

    TT/drop pretty much tracked hit rate here. Remember you can't be talking about fifths, etc. here, just statistical significance. Miner 1 (Leeloo) had both significantly higher hit rate and TT / drop than both miner 2 and 3 (Casper and myself), while both 2 and 3 had significantly lower hit rates and TT / drop than miner 1, but there were no differences in either measure between 2 and 3.

    I did mention this:

    but in order from miner 1 to 3's total TT: Leeloo 33.48, Casper 12.42, and myself 16.55 PED
    Those are just totals and not comparable. I also mention those totals are from non-normal distributions (e.g., skewing), so those shouldn't be used in any comparisons. That's why I used the zero-inflated poisson regression where those averages don't really apply, but gave those numbers with that caution anyways just for a reference point. When I did run those models, I didn't get any evidence contradicting that TT/drop proportionally changed with lower hit rate, but as I mentioned in the OP, presenting those zero-inflated models gets a little trickier compared to the simpler statistics, so I didn't get that far into the nuts and bolts for the purposes of this post.

    That said, I went back into the data, and the back-transformed averages of TT/drop were:
    Leeloo: 0.344
    Casper: 0.120
    myself: 0.080


    If you plot those against hit rate, it's basically a linear decrease in TT as hit rate decreases. Keep in mind that's also accounting for an average across ore and enmatter (i.e., 60 drops per person). Basically, TT doesn't increase or stay flat as hit rate decreases.

  9. #29
    what i'm curious about, did you get some multiplier shortly after the testing?
    the payback hof is a lie

  10. #30
    Mutated mastermesh's Avatar
    Joined
    Apr 2007
    Gender | Ingame
    Male | Female
    Location
    Billy Bar
    Avatar
    Maria Mesh
    Society
    tAS->Entropia Asia Cadets->Entropia Asia->Zoku->NBK->NBK Entrepreneur->Kaos->the Ministry
    Posts
    14,913
    Blog Entries
    10
    Images
    1748
    hmmm... so don't mine where someone else has? Wonder if mining on different planets and 'indoor' might get you different results... Seems like the less popular planets might actually have some use since the population is lower?

Page 3 of 4 FirstFirst 1 2 3 4 LastLast

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •  

Follow Planet Calypso on Twitter  Follow Planet Calypso on Facebook