Life at the end of the curve

This is part rant and part just stomped-into-the-ground discouragement. But it is a dark gloomy rainy day here, and Monday to boot, so….

First a math lesson. (Sorry, I warned you this might be rough.) I am not an expert in statistics, but I understand some of the basics. I know, for example, that if you are averaging the number of widgets per person, and one person has 50 and the other person has 90, the average number of widgets per person is 70. (Statisticians prefer the term “arithmetic mean” but I will use “average” because it sounds a tad less nerdy.) If both have 70, the average is still 70. And if one person has one and the other person has 139, the average is still 70.

In any sufficiently large population, there will be a significant number of actual things counted that deviate a great deal from the average. There’s a term that measures variation from the average, it’s called standard deviation (SD). Basically, the higher the SD, the fewer members of the population that are close to the calculated average. Another way to look at it is the higher the SD, the more spread apart the data is. My example of 1 and 139 would have a very high standard deviation, whereas my example of 70 and 70 would have a very low — zero — SD value.

There is a formula for calculating SD, but as I recall it involves square roots and nobody wants to go there. The point is, it is possible to have an average that no part of the population really experiences. Or that some experience very closely but others experience a long way away from the average. Thing is, in large populations where something is being measured, there are almost always outliers — the ones on the very edges of the curve, about as far away from the calculated average as you can get. Some are way under and some are way over.

When you introduce humans into the math, psychology enters in. If widgets were a good thing to have, and I told you the average number of widgets people win when they roll a die is 70, you would expect to roll the die and win somewhere around 70 widgets, give or take a few. It is just human nature. Even if you fully comprehend how averaging works, you will still expect to get something close to the average. And if you don’t get close the first time, your human brain — against all logic — expects to be “compensated” the second or third or fourth or fifth time. Further, if you roll the die 10 times and end up with 10 widgets, when you were expecting close to 700, and someone else rolls 10 times and ends up with 1000, you will most likely sputter and squeak and even if you don’t say it out loud you will think “No fair! No fair!” And unless you are Mother Theresa you will not be having very charitable thoughts about the other person.

So a few nights ago I was trying — for the third night in a row — to get into just one lousy LFR to crank out my weekly attempt at getting Tomes of Chaos. The first two nights the queue was two hours, and I started playing late so really did not have time to wait that long. The third night night the queue said 35 minutes, so I said what the heck and jumped in. As usual, the queue began to fluctuate, so that when I got to 35 minutes it jumped to 40, when I got to 40, it jumped to an hour, etc. Then when I got to an hour and a half, it jumped back to 40 minutes, which is where it stayed until it finally popped for me after an hour and 50 minutes. I got a total of one Tome out of the raid, which is what I got last week also, so at this rate I only have 31 more weeks to go before I get all my Tomes.

Bad luck, you say. Well yes it is. But I never ever seem to have any good luck, never even any “average” luck, with drop rates or queue lengths or any other activity in this game. I always seem to live on the extreme bad end of the curve. Clearly if the average wait time was 40 minutes, and my wait time was an hour and 50 minutes, then there were some people who had wait times much much shorter than 40 minutes. And they can’t all be tanks and healers, given the proportion of DPS to those roles. If I am going to be a perpetual outlier, it would nice once in awhile to be an outlier on the good end.

Trust me, there are such people, ones who seem always to be the polar opposite of me, ones who get 1000 widgets regularly to my 10. There is someone like that in my guild. Nice guy, good player, but he has the most fantastic luck all the time. Example: I have run Kazzak now 4 times, three times on my main and once on my alt hunter. Rolled bonus roll every time, so I have had 8 shots at gear. Want to know how many pieces of gear I have gotten? Yep, zero. I muttered a bit in guild chat after my last big fat goose egg on him, and this guy — the super lucky one — said yeah that’s tough, but he has gotten the trinket now 3 times and that is no fun either, getting the same piece over and over.

When he said that, I just sat at my computer, and I think smoke was literally coming out my ears. What I said was something like “LOL! Ya that would be annoying.” but what I wanted to do — since I am decidedly not Mother Theresa — was JUMP THROUGH THE INTERNET, GRAB HIM BY THE THROAT, AND CHOKE THE LIVING SH*T OUT OF HIM. Because this guy is an outlier like I am, except he is always on the good end of the curve and I am always on the bad end. Same guy, on a night when I gave up LFR after close to two hours, told me that was strange, since he had queued up about an hour before on his lock and got in after about ten minutes. (!!!!!!!!)

Have I mentioned that I have not got a 4-piece tier set for the last two expansions on either of my hunters? Closest I got was 3 pieces on one in Mists. And that was when I was raiding 3 nights a week with two teams. Back in Mists, I ran Ordos 24 times (multiple alts), had something like 40 rolls including bonus rolls, before I ever got my first piece of gear. I’ve told you about my continuing bad luck with Felblight, which is supposed to have an average drop rate of 10% — mine now after three weeks of 6.2, hundreds of skins, fish and ore, is still right around 2-3%.

I understand the concept of “average”. I know there is no cosmic judge that compensates you for previous bad luck or penalizes you for previous good luck. Even knowing this to be true, though, does not change the human response that chronic bad luck streaks are not fun. Nor does it change the fact that games are supposed to be fun for everyone that chooses to play.

Blizz devs have repeatedly touted that player skill and diligence are the most important factors in this game. Yet what is less dependent on player skill or diligence than strict RNG-based rewards? In fact, no random number generator is truly random, and Blizz tweaks RNG continually. There is no reason, for example, that they cannot tweak their algorithm so that, if the drop rate is supposed to be 1 in 10, it is that rate the first time you try. If you don’t get the drop the first time, the second time you try it is 2 in 10. Then 3 in 10, etc., so that by the time you get to your tenth try — if you have still not got the drop — you have a 100% chance of getting it. That would keep the game fun for people like me. If I knew I could down a certain boss 20 times and get a piece of needed gear, I would do so. Maybe I would get it my first time, but if I didn’t, I would know that if I just kept coming back it would drop. I would not have that sinking, Sisyphean feeling every time I tried.

As I said, it is Monday. It is dark. It is gloomy. It is rainy. And Blizz is Lucy holding the football, and I am Charlie Brown.

About Fiannor
I have a day job but escape by playing WoW. I love playing a hunter, and my Lake Wobegonian goal is to become "above average" at it.

6 Responses to Life at the end of the curve

  1. TheGrumpyElf says:

    Blizzard and math do not get along well. I’ve said it a million times. They do not know how to make random work. Actually, as a person who codes, or at least used to, unless they came up with something new since I stopped coding in the last 10 years then the fact of the matter is there is no such thing as random in terms of coding. It can not be done.

    Not to delve too far into it but certain factors need to be used to create a feeling of random. The first game I ever coded I used the 7th and 8th digit of the moment you logged in, something like 8:34.56781366, the numbers after the . being fractions of seconds of course, to be the randomizing base element. It is what decides your luck for the day you logged in. Using that as a system meant that no one could game it. If I were to use the 66 in that number it would be impossible for your to log in at that exact time to get what might be the “good” generator.

    It was, in my mind, the best way I could come about random generation. Anything else seemed lackluster, too random and completely unfair. Although using that was not perfect I never once had a complaint about bad RNG in the game I made. Admittedly my 700 players was “slightly” fewer than blizzards 7M.

    But that was to explain that random is impossible and my belief that blizzard needs to redesign whatever system to put in place to 1) be more fairly balanced in terms of the standard deviation of luck and 2) to have a true bad luck protection that keeps track of anything that could be considered luck based.

    I personally, if it were me, would move to a reward = effort system (read valor for all gear) and try to remove as much of the random number generation as possible. Broken or not, any RNG system that is intended to be a carrot is a bad idea. People will get frustrated and quit. Carrots need to be attainable, not require luck to get. Random works for “bonus” stuff that people do not consider required. It should never be used for things like gear which many consider required.

    • Fiannor says:

      This seems to be another example of Blizz proclaiming one thing (key to the game is player skill and diligence) but implementing policies completely opposite.

      The gear thing is especially annoying to me. As you say, random — even pseudo-random — should never be used for it, especially for something as key now as tier gear, which at least for hunters seems to be the only thing that brings them into decent balance (MM and BM anyway, not sure anything does it for SV). Not only is that poor class design, it is poor game design, because it is just frustrating and demoralizing to try for it week after week, knowing you could go forever and never see a set, yet maybe someone who has downed the relevant bosses just once can be sporting a 4-pc set.

      I could definitely get behind a system that made all gear valor-earned.

    • Athie says:

      It’s definitely possible to implement true randomness using networked data sources. Here’s a site that feeds truly random numbers based on the quantum fluctuations of the void, for example: https://qrng.anu.edu.au

      For non-cryptographic purposes, this is usually overkill, as pseudorandom numbers are psychologically indistinguishable from physically random numbers — and neither becomes intelligible without hundreds and hundreds of draws.

      Randomness is, for that reason, totally wrong for a game like WoW. Because of lockouts we kill raid bosses maybe up to a few dozen times (farming old content is different). In these small samples randomness is really the same thing as arbitrariness.

      I think we should have guaranteed loot after a certain number of kills with a random chance of getting loot early and a random chance of small but fun upgrades to the loot. That way each kill stays exciting but there is also a sense of clear progress.

  2. Grumsta says:

    I never liked random loot. Back in the paper and dice days of RuneQuest and D&D I hated DMs who dished out loot this way.

    Like those DMs I think Blizz do it for the same reasons: they think it’s “fair”, and they’re too lazy to think further than that.

    Why fair? Because it’s down to the luck of the dice / PRNG. No personal decision so no bias. “Not my fault”. It’s weak.

    As you eloquently put it above, this sucks when you are unlucky. The moment of sweet victory after a boss is killed quickly turns sour when you see the dollop of gold appear, and another dollop of gold on the bonus roll. All that effort for zero meaningful reward. The gold doesn’t even cover the repair bill for the five wipes it took to get this far.

    We’ve debated this endlessly in guild, usually after downing four bosses and all the loot sucks, or it is a duplicate of what was dropped from bonus rolls. The worst example of RNG loot I can recall recently is in a 10 man group we dropped the same pair of chain boots in the group loot, and I dropped the exact same pair in my bonus roll. Whoever wrote the code for that was lazy, following the RNG spec to the letter, or believes that hunter pets need boots too. Who knows. Yes it is of course statistically improbable to get three items all the same, but it wouldn’t be hard to substitute another piece where there’s duplicates. (You could even look at the composition of the party and drop fewer items, but stuff they [gasp] can actually use).

    The solution we all agreed needs bringing in is a token system per boss downed. Harder bosses could be worth more tokens. Keep RNG loot for bonus rolls (with the bad luck mitigator you describe above). That way you trade in the tokens for the pieces you need. Teir pieces and weapons need more tokens (or different tokens). Not hard. (And when you cash the tokens in, the pieces you get have useful secondary stats for your class, not more RNG rubbish). You’d get few pieces overall when you start a new tier (just from bonus rolls), but it’ll quickly even out as the team progresses.

    The only people who will miss out are the previously very lucky ones. And we all hate them, so there’s no downside I can see.

    In the absense of effort or common sense from Blizz, we have mitigated against it in our raid group by implementing group loot with reasonably fair rules for who gets what (we tried personal loot but it doesn’t seem much better than in 6.0 or 6.1). Instead of rolling on each item, it goes to the tanks or healers (in that order) if it helps them, then to DPS if they have the lowest ilvl. Tier pieces go to someone who already has 1 or 3 pieces. In the event of ties, we use a roll.

    The result is folks are less dissatisfied. You may still come out of the raid with no loot, but at least the team is stronger, and getting more bosses down and therefore more chances of loot just got a bit easier.

    Recent team arrivals are an issue of course, and we have given away loot to folks who never turn up again, but that happened with RNG rolls anyway. It’s the best system we can arrive at given the hand Blizz has dealt us.

    • Fiannor says:

      A token system sounds reasonable — almost any non-random system would be better than the current one, in my opinion.

      Your loot system seems to work for your team, so you must be on a relatively stable and responsible team. We tried a similar system on a team I was on, and it failed miserably, mainly because the team composition fluctuated every night. We also found that it rewarded the lazy players who never did anything outside of raid to gear up, and penalized the ones who worked at gearing up outside of raid. Additionally, our system failed to take into consideration nuances of gear. For example, say you had 3 pieces of max crafted gear equipped, and something dropped that was the exact equivalent of one of those. Likely it would go to someone else, even though if you equipped it, it would allow you to go ahead and put a new crafted piece into a needed slot to improve your overall ilevel.

      Glad you have found a way to deal with a bad system, but as you point out, it does not change the fact it is a bad system to begin with.

      • Grumsta says:

        That’s a good point about crafted gear and freeing up slots: I’ll mention that to the RL.

        There was too much whining and “it’s not fair” going on, so the loot system was imposed by the RL with a take-it-or-leave-it ulimatum. It has helped a lot. But you’re right we have a solid core of 14 – 16 players now so it makes sense for our group. Now we’ve tried personal loot no-one wants to go back there, that’s for sure.

        Crafted gear and RNG – there’s a subject I could (but won’t) rant about for several pages. Make an axe,then waste 15 regrinding tokens and still don’t have the Crit & Mastery I need? Thanks Blizz.

        RNG has been thrown into every aspect of this xpac at every conceivable opportunity: even gaining new spells from level 91 to 100 for goodness sake!