Blog Archives

Opaque RNG is Indistinguishable from Rigging

My patience with enforced 50% win rates is paper graphene-thin.

“A fair game is one in which you win half the time.” It’s hard to argue against such a notion. What is more fair than a coin flip? The problem is that players aren’t equal sides of a coin, nor are the thousands of potential actions reducible to two, easily predictive binary outcomes. Some approximation is required. Or a developer thumb on the scale.

I am still playing Clash Royale despite the disastrous pivot towards blood stone squeezing, and the conceptual breakdown of all progression for long-term players. But some of their shit is driving me up a wall, and will eventually drive me from the game entirely. Specifically, Clan Wars, and even more specifically, a particular game mode with preconstructed decks.

To be sure, there are learning curves involved. Supercell basically took some “top decks” and added them to a pool, from which you are randomly assigned one for a single game. The problem is that some of these decks are just objectively terrible with no redeeming qualities, and still others are straight-up countered by some of the other matchups. For example, these two Classic Decks Battles:


Bullshit on the left, winners on the right.

In the first match (at the bottom), my Royal Hogs are immediately countered by Valkyrie, Mega Knight is immediately countered by Inferno Tower. Amusingly, Royal Hogs are also countered by Inferno Tower and Mega Knight by Valkyrie, assuming my opponent times it right. Meanwhile, while I can counter his Goblin Barrel with Arrows, they both cost 3 Elixir and thus end in a wash… with the slightest error on my part resulting in easily >30% tower damage. Meanwhile, my Zappies are basically useless, my Inferno Dragon even more useless, and I can’t use Arrows to counter his Princess or Goblin Gang because then I become vulnerable to Goblin Barrel. I also can’t hope to Fireball him out because he also has Rocket, which deals way more damage than Fireball. The ONLY way anyone could possibly win with the deck I was given was if the opponent was AFK. 1

For a WoW analogy, think Warrior (me) vs Frost Mage (opponent).

The second matchup wasn’t technically as lopsided, but still awful. Bandit is straight-up countered by pretty much every card in the opponent’s deck. Rascals + Zap took care of Minion Horde every time I threw one down, and Hog Rider/Mortar/Goblin Gang meant I could be punished immediately for dropping Elixir Collector or Three Musketeers. Which is what happened, pretty consistently. If I played better, I might have been able to distract a Mortar with my Valkyrie or Bandit in the other lane, and then split a Three Musketeers or something in the middle, followed by a split Minion Horde. Even then, if he played defensive for 20 seconds, my shit would have been countered.

Were these match-ups truly random? Or “enforced” 50% win rates? There is no direct economic incentive for Supercell to “rig” the Classic Decks Battle mode, but the RNG is opaque and it would certainly be a method to ensure that winrates do not get too lopsided.

The third clan war battle I played was Draft. In this game mode, you are given a choice of one of two cards, four times total; whatever you don’t pick goes to your opponent. I’m not sure if the card pairings are 100% random, but you can absolutely get stuck with some extremely shitty decks and/or matchups. And yet I’m fine with that. You as the player have some agency, even with imperfect information, e.g. choosing Minion Horde when opponent might have chosen Arrows. Indeed, Minion Horde in particular is a classical risky pick because of how many cards can counter it… but if your opponent doesn’t have any of those counters, it can be an overwhelming advantage.

My feelings on enforced winrates have changed over the years. Initially, it seemed fine. Necessary, even. But it is rigging, especially in the methods that many game developers go about it: pairing you with terrible teammates, matching you against strong counters, etc. The end result is that I simply cannot trust game developers with (opaque) RNG anymore. They have no incentive to be actually fair – however fairness is defined – and every incentive to produce favorable (to the devs) results. Even if they showed me the specific game code that chooses the matches, I have no reason to believe it operates in that way. This age of monetization and consumer surplus erosion has pushed me past the Cynicism Horizon, from which no trust can escape.

The only thing that game designers can do, and the thing they should be doing, is increasing player agency in the RNG elements. Drafting feels fair, even when the results are not. Maybe it is just another psychological trick to employ, giving someone the “choice” between a rock or a hard place. But it is an important one for not appearing so nakedly rigged in favor of one particular outcome.

1 If you can produce some videos of pros beating non-AFK people with the decks I was given, I’ll concede that I need to L2P. I typically end the season at 4800 trophies and can acknowledge mistakes, but on paper and in practice, those match-ups felt lopsided as hell.

Surplus Deficit

Last week, Keen blogged about a tweet that should be filed under “Things that make you go Hmm… not really”:

In a world of $5 lattes a game with 50 hours of content is worth $1,000. Instead, many won’t touch a game until some stupid Steam Sale. (source)

Wilhelm has already penned an exceptionally good take-down of the latte vs game comparison. What struck a cord in me the most, though, was this follow-up tweet:

The unwillingness to pay what a game is actually worth is why we have on disc DLC, F2P, micros for single player games, season passes, etc. (source)

This, my friends, is the embodiment of everything I warned about six years ago.


We as consumers have been beaten down so often and for so long that the argument almost makes sense. It seems “fair” that someone gets paid a proportional amount for the benefit received. But the funny thing is that reasoning only ever seems to go in one direction. Price exceeds the amount it costs to create? Capitalism, working as intended. Benefit exceeds the price? Suddenly there’s a whole lot of hand-wringing and articles about Millennials killing functionally useless industries.

Fight for your own Consumer Surplus! The difference between how much you paid for something and the amount of enjoyment it provided is yours. That’s your profit, not the game company’s. These corporations will try to erode your consumer surplus with ever more novel monetization schemes, and other people might try to guilt you into “supporting the devs” or admonish your “unwillingness” to throw your hard-earned money in a hole for literally no reason. But the fact remains that it’s the game company‘s responsibility to effectively manage their own resources, to figure out what payment models they should utilize, etc. Not yours. Their business is not your responsibility.

Don’t settle for the precise intersection between Supply and Demand. Don’t let anyone make you feel guilty for getting a deal. If you want to donate extra money to random devs in some idealistic hope they generate future value, go for it. But understand this: the only person looking out for you, is you.

OT: LFD and Difficulty

OT = Off-topic, e.g. no AH advice.

Chances are good that you have at some point been exposed to the debate still consuming the WoW forums in regards to the recent nerfs to heroic difficulty, the buff to Luck of the Draw, and the overall “Wrathification” of Cataclysm. The arguments are pretty rote by this time, usually coming down to “morons/bads should L2P and not have faceroll epixs” and “back in my day we wiped and liked it” and possibly “the elitist no-lifers just want exclusive content for themselves” or “my $15/month is just as valuable as your $15/month.”

All of that kind of debate is besides the point.

The point is two-fold: the DPS queue for heroics is north of 45 minutes and the completion rate of LFD groups was garbage. Period.

LFD is Here to Stay
It is interesting from a philosophical point to debate the whys and the hows, but again, it would be besides the point. The LFD genie is out of the bottle, and it is never going back. That said, LFD as a tool requires a healthy feedback loop in order to function. Players like Gevlon from Greedy Goblin might refuse to use LFD under any circumstances, even if that meant he simply was never able to do another heroic again. I would imagine that everyone else would be more reasonable insofar as they would prefer grouping with friends, but if they could just grab someone to fill the empty slot(s) they would. Sometimes only three people you know are online, or perhaps only three out of X many are willing to go. Other times you may literally be the only person online for whatever reason and want to run a heroic. In those situations, you will want the system to be there.

LFD in this sense is like public transportation. You may never actually need to use it, and you may certainly never want to, but it is still in your best interests for it to be there in case you do.

That all changes if the average completion rate of LFD pugs is 40% after having waiting for nearly an hour. Most sensible people would not bother with that, and instead take their chances with Trade chat pugs, waiting for guildies to become available, or simply going and doing something else entirely. The people who would still queue for such a LFD failure would be the terminally optimistic and those for whom a 40% chance of success after an hour of waiting is more than they achieve on average anyway. This means that when you end up needing to use LFD to avoid not being able to do what you want to do, you are far more likely to not end up being able to do what you wanted to do anyway and wasted your time besides.

What Does Not Wipe You, Makes You… Err
Were the Cataclysm heroics too hard? Would the “difficulty” have solved itself once tier-gear was available for Justice Points in 4.1 and beyond? Is having players struggle through difficult content better for them and the game overall? Interesting questions… but irrelevant to the issue of the negative feedback loop the LFD tool was stuck on.

Success breeds success. There are highly successful people IRL who say that success is the worst teacher, that adversity and frustration are better motivators. Sure… sorta. I am not opposed to difficult content – fundamentally I believe everyone who plays wants content tailored to their skill level – what I am opposed to is the notion that a LFD system could survive the same design philosophy used in, say, raiding. If your first few forays into eBay or Craigslist ended up in scams, frauds and embarassment, how likely would it be that enough people would trooper on in the face of such adversity to make those marketplaces function on a healthy level? Even on a raiding level, success breeds success. How long would Gevlon’s experiement have lasted if endlessly wiped on Magmaw and saw no improvement from week to week?

Does the 15% LFD buff and targeted boss nerfs make people better players? Not necessarily at first. What those things do accomplish is increase the completion rate of LFD dungeon groups as a whole, which then encourages more people to use the tool, which improves the aggregate skill level of groups, which further increases the completion rates. I truly do believe that a smoothly functioning LFD tool encourages individual improvement in the people receptive to the idea to begin with, as they get a foundation of success that translates into confidence, plus the gear that takes raiding into the realm of possibility. Just think of how many potential raiders could be buried under the fail of current LFD groups, never knowing how much better they could become because any improvements they do accomplish does not translate into meaningful group success.

Collateral Damage
The final thing I wanted to briefly talk about is the following argument:

“This sort of thing is exactly WHY the 15% was bought in.
Are you saying we should all fail just because some of the peeps were scrubs?”

“No they should not reward poor play. Groups should fail sometimes people should learn mechanics. The issue is they put the bar on the floor for wrath.” 

One of the root design questions of LFD is: should good players be penalized for getting randomly grouped with bad players?

There is no way to avoid rewarding poor play without also penalizing good play in the process (unless all five players are bad). Such a philosophical hardline is the same concept as the teacher punishing the entire class because no one came forward to say who threw the spitball or whatever. The idea is that by holding the entire class hostage, one can guilt either the perpetrator or someone who knows who the perp is into confessing. Based on personal anecdotes, such a gambit works approximately 0% of the time – someone with no compunction against being disruptive in class in the first place isn’t likely to be persuaded by guilt (even if they are, they are getting punished either way, so no-win), the people who knew who did it likely don’t want to be labeled as snitches and otherwise suffer retaliation later (no-win), and the rest of the class that would tell who did it if they knew are punished as though they did it themselves (no-win). Such blind, indiscriminate punishment does not actually encourage any good behavior whatsoever; the only real thing it encourages is either acting out yourself (may as well have fun if you suffer the consequences either way) or an avoidance of that class/teacher, which represents the LFD tool in this case.

Is it “fair” that bad players get carried? Maybe not. Then again, I’d say the downside of being bad is being bad. If someone is so conceited and ignorant that they are unable to recognize their own terribleness, they are not likely to learn anything from the group being wiped either. Meanwhile, I do not think anyone believe it fair that otherwise good players get punished for something they had no hand in doing.

[Should good players be penalized because of bad players?] In a word? Yes.

Because the flip side of the question is, “Should a player be guaranteed a successful run no matter what the other 4 people in the group do?” And the answer to that is, “No.”

A good point, but presumably the line does not exist at such an extreme. Should two good players be punished because three bads happen to be in the group? How about three good, two bad? Four good, one bad? And what about when the the binary distinctions are dropped, and we start adding “above average” “average,” and “below average,” to the mix?

Ultimately, I believe the changes which Blizzard did were amazingly nuanced. The targeted nerfs are nerfs, of course. But full guild groups looking for challenge can avoid the 10% portion of the buff by doing what they always did anyway: run heroics as groups. And if they ever need a 4th or 5th member to round out the run? At least they will not have their run torpedoed by a kid shooting spitballs.