Jump to content

Yuzral

Members
  • Content Сount

    267
  • Joined

  • Last visited

  • Battles

    284

Community Reputation

92 Good

About Yuzral

  • Rank
    Master Chief Petty Officer
  • Insignia

Recent Profile Visitors

971 profile views
  1. First off: I am aware that the only suggestion a vocal chunk of the community will entertain for subs is that the concept to be towed out to the Laurentian Abyss, submerged beneath them and then scuttled1. Accepting that - and that it's not likely to happen - I'd like to float the idea of adding audio cues for submarines in the various detection/near detection states. Specifically the iconic 'ping' of a hostile sonar hitting the boat's hull. On a practical level, this gives sub commanders an audio cue that they should be deep and staying there, as well as adding a bit more immersion and maybe making it that little bit more tense for them. Which is my excuse for the pitch of the ping being at a frequency that wasn't really used by WW2 sonars (as I understand it), changing the pitch of the ping depending on the sub's detected state and not caring whether or not the detecting vessel actually had such a system, especially at the sub-2km range: So - if the sub is below periscope depth and the following conditions are true: "Watchful" skill in effect - Sub player hears a high A/A5/880Hz ping, repeating every 5 seconds. Probably not your most immediate problem, but pay attention. Detected by hydro search - Sub player hears a 1760Hz/A6, ping, repeating every 2.5 seconds. You should be doing something about this. Proximity detected below periscope depth - Sub player hears a 1760Hz/A6 ping, repeating every 1 second. You can panic now. Thoughts? 1: And make sure the count is accurate this time, doctor.
  2. Yuzral

    Sad State of Affairs ..(Streaming)

    Are you looking at YouTube? While a lot of the EU streamers are offline right now because it's somewhere between 1 and 4 am for us, the category page for WoWs on Twitch has Stats on 480 viewers, Bou's got 310 while he bimbles around on the PTS, Starlord on 101 and that's just the top 3.
  3. Fixed, thanks. If one typo is the worst sin in a post I finished on the wrong side of 1am then I’m a happy man…
  4. So, here we are after just over two weeks and I'm in the weird position of having to declare "task failed successfully". The TLDR is this: The data gathered does not reliably confirm or deny the hypothesis that the Missouri was any more or less likely to occur than any of the other 40 possible outcomes. However, WG (in the shape of Ev1n) have stated that there was an equal 1/41 chance for each outcome, which as far as I'm aware, is the first time WoWs-PC staff have ever officially stated a drop rate. With that out of the way, it's maths time. I'll start with the modelling I did two weeks ago. Specifically, the equation p(M)n = p(M)1/[p(M)1+41-n(p(!M)1]. Yes, I realise that looks like a faceslam into the keyboard. Breaking it down: p(M)n is the chance of getting the Missouri on the nth draw from the deck. p(!M)1 is the probability of a given not-Missouri card being drawn from the deck on the first draw - I'm assuming that the other 40 cards all have an equal chance of being pulled. p(M)1 is the chance of getting the Missouri on draw 1 and n is the draw number, starting at 1 and ending at 41. How did I get there? I started with the 'even odds' case: 41 cards in the deck - 1 Missouri card and 40 not-Missouri cards with each outcome having the same chance and if the Missouri card isn't in the deck, we aren't interested. First draw - 1 Mo, 40 non-Mo. Second draw - 1 Mo, 39 non-Mo. Third draw - 1 Mo, 38 non-Mo. Generalising that, there will always be one Missouri card out of a total of 1+(41-n) cards in the deck on the nth draw. Basic probability kicks in and tells us that the odds of drawing the Missouri will be 1/[1+(41-n)]. As a check, the 41st draw would be 1/[1+(41-41)] = 1/1 = 1 = certainty because the Mo card would be the only one in the deck at that point. Now let's generalise a bit. In the example above, I counted ‘cards’, but I can also express the situation as a sum of probabilities. For the first draw, I have one event with a probability of p(M)1 and 40 events with a p(!M)1. Incidentally, since this covers all possible outcomes in the initial deck, p(M)1 + 40 x p(!M)1 must equal 1. Therefore p(!M)1 = [1-p(M)1]/40. Critically, these individual outcome chances are fixed, even as the size of the deck decreases - the changing odds at each draw are reflected by the change in n. You can probably see where this is going: For the first draw I have 1 p(M)1 event and 40 p(!M)1 events. For the second, 1 p(M)1 event and 39 p(!M)1 events....and so on. For the nth draw, I will have 1 p(M)1 outcome and 41-n p(!M)1 events and this gets us to the key equation - p(M)n = p(M)1/[p(M)1+41-n(p(!M)1]. Since I could now calculate the probability of each individual draw getting the Missouri (even with a stacked deck), I can also calculate the probability of each individual draw losing and thus the actual chance of getting the Missouri at any given draw since to get Missouri on draw n, you would have to not get Missouri on all the draws prior to n. So for the actual chance of getting Missouri on, say, draw 5, multiply the individual chances of losing draws 1, 2, 3 and 4 and the individual chance of winning draw 5. This all went into a Google sheet, which I will not be sharing until I figure out how to strip my private Google ID out of it. I have no interest in doxxing myself. Suffice to say that it worked. For a 'fair' p(M)1 of 1/41, a graph of the cumulative chance of getting the Missouri on draw n versus n returns a straight line, indicating that the chance of getting the Mo was the same at any individual step. That's the blue line in the graph below. The red line is what happened when I dropped p(M)1 to 1/50. And yes, this particular bit of shenanigans would have seen WG taking an average of 6,000 doubloons extra per player. This is where I was just over two weeks ago: I had a model that seemed to work and predicted what the end result would look like depending on what chance I gave the Missouri card of being drawn. Now all I needed to know when people were getting that Missouri card. Lots of people. This is where things fell apart. To be reasonably confident, I wanted an average of 15 results per pot. Problem was, with 41 pots, that meant needing 615 reports of getting a Missouri. I got a total of 96 from all sources. To further compound things, Duckyshot made the point that the study would tend to bias towards a low n simply because a number of people would give up before reaching high values of n. As a result, I have to declare failure on the primary objective - I am woefully short of the results needed to have any real confidence and there's at least one flaw in the study which I'm not sure I can correct for (maybe by comparing with the populations 'tapping out' at each stage?). That said, a few odds and ends were salvageable: 1) Ev1n confirmed on the US forums that the Missouri 'card' did have the same chance of being drawn as any other card. 2) It is worth observing that of the 96 Missouri cards, 77-80 came from the NA community. Either people in the EU are a lot less likely to engage with this kind of event or they're a lot less likely to talk about it to a random Internet stranger even when given the option to do so via private message. 3) "First result bias" is a thing, to a tune of about 300% over-reporting - probably because the first card was free to look at and therefore having a much larger population looking at it. Despite supposedly having the same 1/41 chance of being the Missouri as any other point in the draw, 16 of the 96 results reported were n=1 draws of the Missouri. Of course 16 of the 32 people who reported not getting Missouri tapped out on the first draw and a separate survey was, I believe, suggesting about the expected 1/41 chance.
  5. Yuzral

    Important message for the community

    1) I beg to differ. That I’d have to put any effort in to ignore it is a good illustration of how angry the last few weeks have left me. In a rather repressed, stiff-upper-lip English fashion, of course. 2) The only message you should be getting from that is that I forgot to remove the clan tag before I uninstalled the client. But it serves as a useful marker that I’m actually the Yuzral who bailed and not some yahoo so it does still serve a purpose.
  6. Yuzral

    Important message for the community

    If that request landed in my inbox tomorrow, the reply - once I’d edited out various obscenities that part of me would want to add - would be ‘No’ and I suspect the same is true of the other CCs who left. Wargaming will need to deliver on these promises and do so consistently over an extended period before I (and maybe the other ex-CCs?) start to trust them again. Edit: And on the topic of China, aren’t they on their own code branch and running through a partner company? Vaguely recall that they had RTS CVs long after the rework…
  7. SeaShadow - My assumption was B. Of course, assume makes an [donkey] out of u and me, hence this whole exercise. Edit: American automod. Of course. See what I mean about assume?
  8. Dropping a marker for my own use here - results to this point have been collated. Thank you to the 60 people who responded - now I need to go and get the EU's data and see what the first day's results look like. @Ev1n_NA - Didn't expect you to swing by! Thank you for the data and also the statement that each 'card' has the same base chance of dropping. As a bit of preliminary feedback: That level of disclosure is extremely welcome. Please consider similarly disclosing odds for decks like this and WG's other gambling and quasi-gambling promotions in the future. While you'd still have to deal with varying flavours of anti-gambling opinion and the "WG are lying liars who lie" crowd, people would at least be fully informed.
  9. This is never going to be anything more than a preliminary study with a decent helping of uncertainty fudge. Even before you start trying to allow for incorrect or downright dishonest reporting, the only way I have of reaching enough people (namely, forums and socials) is still going to miss probably 90%+ of the WoWs-playing population and worse, may inadvertantly select for extreme results. People like to show off the 1/41 or 41/41 results but a 22/41? Eh. Hopefully a decent size sample will smooth that out to an extent but a comprehensive audit is well beyond my resources.
  10. Good morning everyone! Last night, I and a few of the other ex-CCs were talking. Some mathematics may have happened and it ended with LittleWhiteMouse pointing out that it would need 600+ results to get a decent data set on the odds of Missouri dropping. This means I need to crowdsource my data and that means I need your help. Specifically, if you have gambled for and got the Missouri, I want to know how many boxes/rolls you used to get it. Please note - if you weren't planning to gamble for Missouri, don't let this change your mind. Edit: If you’re not happy publically revealing how much you spent, then please feel free to PM me the number instead.
  11. Small detail: Turry, not Tuccy. Different people. Unless something bad happened to Tuccy that I don’t know about…? For the record, I’ve had one DM on Discord from Sub_Octavian asking if I’d sent a copy of my resignation to Shonai…which was in response to my DMing a copy of said resignation to him. When Mademoisail claimed they’d reached out to the ex-CCs, she was using ‘CC’ as a collective noun and could more accurately have said ‘a couple of the ex-CCs’.
  12. A quick and rather unscientific test (haha to the 1+1=4 post and cool to the bad cop post) had me able to react to a couple of Chobittsu’s posts without incident. Whisper it…but that might really have been temporary gremlins.
  13. Yuzral

    They found their scapegoat...

    Because the offending code was known about globally and caused a global PR whoopsie, therefore the apology needs to be global as well? That would be my argument for such a decision, although personally I’d have kept the code RU only as well since that’s where the first worked. That said, per Yabbacoe on the EU boards, TURRYWEARESORRY should now work globally.
  14. Huh. I remember wondering at the time why WV wasn’t called Maryland and the WV name kept for a possible 1944 refit. And just how much the hype over Fallout 76 at the time had to do with it. Didn’t know there’d been a commitment on the point. Getting back on topic: 1) Given the last 10 days, is WG going to reconsider yet again and simply put the Missouri on sale for a fixed price without preconditions, catches or miscellaneous shenanigans? 2) Given that the likely answer to (1) is ‘no’, what details can you share about the Missouri campaign missions? 3) The increasing prevalence of gambling and quasi-gambling in WoWs has caused you significant PR difficulties, both in the last few weeks and - at a lower level - long term. Are there any plans to reduce such elements or provide transparent fixed-cost alternatives to them? 4) How much of a say do the regional offices actually get on monetisation, sales, policies and similar? Are things decided locally or do you simply get told what to sell by St Petersburg? And, as tradition demands: 5) Kitakami, when?
×