Loot Boxes

Shoryuken.com asked me about loot boxes and published this article.

For anyone wanting more context, here is the complete set of questions they had about loot boxes in Fantasy Strike and beyond, and the answers I provided:

SRK Questions

  • Any idea when this feature will be implemented?
  • What can players expect to find? Costumes and cosmetics, right? Will there be more rare ones than others? Will there be things like victory poses, emotes, etc. ?
  • Anything that affects gameplay? Or any other effects such as XP boosts? For example, a boost for Arena Runs?
  • Will there be an option to use money to buy more loot boxes?
  • There's a lot of controversy lately in the games industry when it comes to the philosophy behind loot boxes. What are the thoughts on the ongoing discussions and will Fantasy Strike have any specific strategies to not alienate gamers?

Sirlin's Response

It’s unlikely that Fantasy Strike will have loot boxes simply because we now lack the budget to properly implement the feature and all the associated content. But to answer your questions…

Loot boxes were the plan since as long as I can remember. Our site went public in November 2016 and had all the same info then as it does now.

Loot boxes were to be ONLY cosmetic. Period. Costume colors, alternate animations for non-gameplay stuff like victory poses, etc. No gameplay-affecting things.

There has been controversy of loot boxes lately, yes. There is a wide range of implementation about what “loot box” means. I’d like to make it clear that at no point ever, not even for one millisecond, were we ever planning to implement any kind of pay-to-win thing. In my opinion, any supposedly competitive game that allows players to pay for power or grind for power is doing a great disservice to fairness in competition, which I think is an important value. I don’t play or support any such games. To me, THAT is what the real controversy should be, as it includes business models from big games out there, often not even using loot boxes, that ruin the concept of an even playfield in competition.

Back to loot boxes, we were planning (but may now not be able to implement) them in the vein of Overwatch. That is, cosmetic only, players would get them for free over time or could pay for me. It’s not like Overwatch invented this idea, but it’s a high profile example, and one I followed very closely. Blizzard also makes Heroes of the Storm, a game that sold cosmetics directly, often for $10 each (I’m aware they switched to the Overwatch system now). I wondered, at the time, what the player reaction would be to Overwatch’s different system. So I read everything I could from players. And I do mean quite literally that I read multiple thousands of comments directly from players. Every thread on every gaming site I could find, even some hundreds of pages long.

The reaction was quite clear. I found something like 3 posts that were upset, in total. And thousands all in favor. It really was that overwhelmingly one-sided. The reason is simple: in the Heroes of the Storm way, most players felt like they’d simply never own any of those cosmetics. In the Overwatch lootbox way, players get stuff for free. And they really really like that. Not only that, but they have the potential to get any (and all) of those cosmetics for free, meaning that there isn’t any class of items withheld from the boxes. So this is the spirit we were operating in, with our plans. The thing where we get to give away stuff for free and make a large number of people happy.

The positives don’t stop there. Another thing true of Overwatch’s system, which is like 100x more true for us than them, is that this ongoing revenue stream is something that raises all boats for the playerbase. Every player benefits because the ongoing revenue stream allows the developer to pay for continued development, more free features and free gameplay content for ALL players. In Overwatch’s case, it means you know that new characters, maps, and game modes can be released for free, and they have been. I think it’s incredible (in a good way!) that they were able to give all players access to all characters, even new characters, without having anyone pay for an expansion or pay for each new character. It’s a case where the theory really did work out in practice, in that everyone benefited by getting free gameplay-content in addition to the part where they got some free cosmetics too.

I said the above would be much more true for us than Blizzard. I don’t mean that Blizzard is somehow lying about the money involved. All I mean is that Overwatch in particular happened to be so wildly successful (half a billion dollars in revenue? Or a full billion by now?) that they could in theory decide to fund continued development anyway. Now, they might decide to really scale back on that if they didn’t have the ongoing revenue, but that’s a choice that would be up to them. It’s not a choice that’s up to us though. If we don’t have continued revenue from something like loot boxes, there is no way in the world we, as a struggling indie, can continue to pay for our team to develop new free gameplay for everyone. Like any normal company, we’d have to choose between no further content, or paid gameplay content.

I’ve mentioned the ways in which loot boxes are actually good for players (free cosmetics, and free gameplay stuff too for everyone) and developers (can afford to continue development of free stuff for everyone). The recent controversy of it wall was sparked by much worse business practices than this. The whole thing about $80 for a Star Wars game then you still have to grind 40 hours to get Darth Vader really made everyone mad, and rightfully so. And let’s imagine that this made lots of ongoing revenue for EA. Does anyone really believe that the result is going to be a bunch of free gameplay content like new characters and maps, the way Blizzard would do it? I don’t think so. That sounds very not-EA to me. I don’t have much specific insight into this situation with EA games, but my sense of it, and probably the sense of most gamers on the internet, is that EA really wants to stick it to you the maximum amount, period.

I think that even though both examples I’ve talked about are “loot boxes,” that there are really opposite mentalities behind them. One that is very much trying to help players. To give them some free cosmetics instead of none. And more importantly, to fund free gameplay content. Overwatch releases new characters that are completely free. As in, you instantly get them, no paying or grinding is involved. Can you imagine if Street Fighter did that? Or if Fantasy Strike did?

The opposite mentality is this kind of thing basically used for evil. I’ll ask again, “Can you imagine” if EA used the money raised from loot boxes to put out new free content like a new character? The notion is laughable--I’m actually laughing right now just thinking about it--because they already went as far as possible from making Darth Vader free. I think gamers really *get* that. They get that there’s something really cynical, cash-grabby, and directly contrary to their interests going on there. Yeah that’s probably true in EA’s case. But I think Overwatch has shown that a non-evil implementation is a huge boon for players. And that’s the type of thinking we had when considering loot boxes--the intention of using that revenue to fund further gameplay development that ALL our players would benefit from.

Adding a Friend in Street Fighter 5

I was thinking about how we'll handle cross-platform play in Fantasy Strike. The plan is that you'll add a friend to your in-game friends list, even if they are on another platform, then you can click on them there to directly challenge them.

I was looking into how Street Fighter 5 handles your in-game friends list. They went a different way with it. Below is a log of my experience.

Sirlin: I'd like to give everyone some unsolicited advice on setting up your SF5 friends list.

Sotek: ok sirlin, what is the advise

Sirlin: You might want to add someone as your friend in SF5. Or maybe you should put them on your favorites list? Or maybe those are the same thing? No, because favorites was in since launch but friends are new. So how should you start?

Sirlin: In the game, you go to CFN, then the submenu has choices: blacklist, Pending CFN Friend Requests, Favorites, Replays, Rival Search, Replay Search, and Ranking.

Sotek: CFN

Sirlin: Keep in mind you can't read those choices all at once. You can only read them one at a time as you scroll through a horizontal list.

Sotek: Capcom Friend Network?

Sirlin: Capcom Fighters Network

Sirlin: "Pending CFN Friend Requests" sounds like it has potential. Maybe you can add a friend there?

Sirlin: No.

Sotek: obv not, you accept requests there

zem: cfn is the united id service since there’s cross platform play

Sirlin: Favorites sounds like even more potential. Maybe you can add a favorite there. Or a friend?

Sirlin: No and no.

zem: i do remember seeing the friends list added and not finding the option to uh.. use it

Sirlin: Ok, "Rival Search" then?

Sotek: I enjoy this "advice"

Sirlin: You can choose from Refine, Search Using Fighter ID, or Search Using Steam ID. We're on the right track folks. Let's choose Search Using Fighter ID, type it in, hit enter, and see what we get.

Sirlin: A list of users! Including the one searched for and others with names containing that string too. Ok click on the one you want.

Sirlin: Oh sorry, mouse cannot be used anywhere in the menus. So press accept on the one you want, which is the B key, told to you no where. But you already had to know that to get anywhere near this far.

Sotek: great

Sirlin: Doing this brings up a "details menu" with exactly two items (not that detailed?). "View Fighter Profile" and "Add to Favorites"

Sirlin: damn it, I wanted to add them as a friend? Is that like not possible? It's really baiting to me add as Favorite. If you add as Favorite, then go to the Favorites menu and see them there, can you THEN add them as a friend?

Sirlin: No.

zem: lol

zem: will we ever find the option to add as friend?? i’m on the edge of my seat

Sirlin: The favorites thing was a decoy. What you really want to do in the "Details menu" of two items, is choose the OTHER item, which is "View Fighter Profile"

zem: ah.. should’ve known

Sirlin: Now we pause for a moment since the game crashed for the 2nd time in the last 10 minutes as I idle in the menus

zem: ahaha
Rexford: The mix-ups????
Sotek: ahahaha
Sotek: the shocking twists, lol

zem: that’s the game’s version of random super (or chaos super, my preferred term)

Sirlin: Ok, we are now "Viewing Fighter Profile". It's a screen full of junk. The most prominent thing taking up most of the screen is a radar graph of how much arm wrestling, vs olympic 1st 2nd 3rd vs heart that player has. With the radar graph completely empty and non-functional of course.

zem: when it builds enough meter (memory leaks) it can play super (crash at any point)

Sotek: arm wrestling vs podium vs heart

Sirlin: There are 5 tabs of info, navigable only by LB and RB (good luck finding those on keyboard)

Sirlin: Damn, which of these tabs should we go to?

Sirlin: None of them.

Sotek: !!

zem: yes, there’s a sub menu!

zem: of course!

Sirlin: In the corner, outside of all these stats, is small help text for "Y: Sub Menu"

Sirlin: Don't actually press the Y key though.

zem: lol

Sotek: wow I was just defeated

Sirlin: Press the keyboard key that corresponds to Y, which is actuallh H.

Sotek: oh ok
zem: was that an intentional typo because it works perfectly
Sotek: lol that typo
GRAG∶ The Translations Guy™: :100:

Sirlin: This brings up a sub menu that looks like the "Details List" except it's not called anything this time. You can: Compare, Add to Favorites, Send CFN Friend Request (!!), Add to Blacklist.

Sirlin: We did it folks, we found friend request.

Sotek: yay!

Sirlin: But what will this actually do once accepted? We already have a favorites list. So is there a favorites list AND a friends list??

Sirlin: Any guesses on that

Polari: I love the stair climbing ability stat

Polari: can't wait for it to be implemented for real [Editor's note: this chat line was written 22 months after the game launched.]

Sotek: I guess: there's a friends list somewhere separate from a favorites list

Sirlin: Before we discover the answer to that, what is a "favorite" anyway?

Sirlin: Someone in your favorites, you can stalk them and determine when they are online / offline, what they are doing online, and when they last played. (Hmm, seems like this should have required their permission??)

zem: my opinion is that favorites was what they implemented before they finished their (clearly well thought out) friends feature

Sirlin: that's so much stuff, maybe we don't need them as a friend? Oh, but what about actually playing against them. How do you challenge a friend to a match from the favorites list?

zem: i have a bonus twist to this story that’s ps4 only

Sirlin: er, challenge a person on your favorites. Not a "friend", because you aren't friends in this example.

Sirlin: The answer to that one is...

Sirlin: you can't.

Sirlin: You can't challenge people on your favorites list to a match. To play them you create a lobby somewhere and invite them. I won't detail that, I assume it's some other whole excruciating process though.

zem: my twist comes in during that process

zem: it is

Sirlin: If anything, that makes the friend feature more exciting though. Because you need to be able to really challenge people and favorites can't do it.

Sirlin: Ok, so you send the CFN Friend Request, they accept, and now you're friends! Awesome, lets' check that friends list.

Sirlin: Oh...there isn't a friends list. There's still just a favorites list? It adds them to that list, exactly like anyone else. The "friend request" just adds you to their favorites too, instead of only adding them to yours.

zem: loll

Sirlin: So you still can't challenge them directly.

Sirlin: The end.

Sotek: Surprise ending

zem: here's the ps4 twist: if you're streaming using the ps4 OS feature, there's no way to send or receive an invitation without halting the stream. opening up your favorites to invite someone to your lobby halts the stream. opening up your "system messages" to accept someone's invitation halts the stream. no other way to play with friends as far as i know

Polari: the friends list was dead all along

Sirlin: Yeah I noticed that. I guess they are trying to prevent streaming of everyone in your friends (er...favorites) list? Like a security thing?

zem: apparently! or your important messages, which include invitations from people who might know you, and system messages about getting fight money or downloading updates

zem: at least the buttons are labeled right though!

Sirlin: I was going to close SF5, but it just crashed for the 3rd time and closed for me.

Sirlin on Game Design, Ep 18: Fantasy Strike

Fantasy Strike is our new fighting game. We've just announced it and begun crowdfunding for it on Patreon. This podcast explains what Fantasy Strike is all about. We cover the high concept about making a fighting game more accessible from top to bottom than anything else we've seen, the specific game mechanics we chose to accomplish that, and the resulting dynamics of how it plays.

Because it's unusual to crowdfund a game through Patreon, we also explain why we're doing that and what the advantages are.

Hosts: David Sirlin, Richard "Leontes" Lopez, and Sean "MrGPhantome" Washington.

Sirlin on Game Design, Ep 17: Shipping Kickstarters On Time

Kickstarters projects are notoriously late. I give advice on how to ship your Kickstarter project on time. I've shipped 5 out of 5 Kickstarters on time, so it's time to share the best practices of how you can do that too. Learn about how complete your board game should be before you take it to Kickstarter, about the "magic word," and about shipping shipping shipping!

While most of the episode is about Kickstarters for board games, later in the episode we cover the extra challenges of doing a Kickstarter for a video game. And we reveal that we're going a different route with our Fantasy Strike video game, which is using Patreon for crowdfunding right now.

Hosts: David Sirlin and Richard "Leontes" Lopez

Fantasy Strike patch notes, August 2016

Thank you to all our patrons for supporting the Fantasy Strike fighting game. Although we're not ready to show the game publicly yet, we give our $25+ patrons new builds every month, the latest one is here. We rely on your support, and we'll need even more of it to finish the game. If a really accessible, easy-to-play fighting game that's still tournament-quality appeals to you, please support us.

If you need more convincing, no problem. In a few months we'll finally show the game publicly and give you more ways to help us fund the game to completion. For those who are interested right NOW, here's what's in our latest build.

Change List:

The biggest thing is a new playable character: Midori. That brings us to EIGHT playable characters now. Also worth noting are lots of new animations for Valerie, a new character model for Grave that has over 3x the polygons as before, and a preview of a new character select screen that will be used for modes other than local versus and training.


--New character!
--Midori’s super takes a long time to build. Using it lets him transform into a dragon. You can build super faster by using his parry (ground C).
--Normal attacks in human form. b+A is a sweep, A is a chop that can be cancelled to specials, and f+A is a 2-hit attack.
--B is a torpedo attack, can also be done in the air.
--C is a parry. If an attack would hit you as you parry, you take 0 damage from it and automatically hit back. In addition to dealing damage (if you hit back), you also get super meter, AND you gain an empowered state where you glow purple. In this mode, you will AUTOMATICALLY throw the opponent if you’re in neutral state and they’re throwable. You lose this state if you use it (to throw them) or if you get hit. This state doesn’t persist across rounds.
--Air A is a kick.
--Air C is a flying butt stomp thing. Crossups!
--Both ground and air super transform you into a dragon.


--Oh my!
--The dragon’s character model is an unfinished, low polygon version. Also, the dragon has many glitches that we know about, especially related to throws. Because he is generally playable, we wanted to get this to you before fixing his various animation problems.
--b+A is a sweep with absurd range.
--A is a an attack that hits in front and above. You can press A,A to get a second attack that flies up into the air. After that, you are still able to do another air attack before you land.
--f+A is a pretty long range attack that you can cancel to an acid spit as a second attack by pressing A again.
--B is like a better version of the human version’s torpedo attack. On hit, it bounces back and you can do another air action before you land from the bounce.
--C is a running grab with 2 hits of super armor.
--Air A hits twice and is crazy.
--Air B is an air version of ground B; a torpedo attack.
--Air C is as a divebomb that grabs opponents. Can be yomi countered.
--Normal throws do 2 damage in Dragon form.
--Your super meter becomes a countdown meter in dragon from. You automatically revert to human Midori when the meter runs out.
--New debug option in training mode to stay in dragon form forever once you enter it.


--New character model. Grave now has over 3x the polygons he had before. Also, his hands have been rebuilt to be better shaped and allow for more articulation.
--Grave’s costume colors now use our new system (already in use by Setsuki, Geiger, DeGrey, and Midori). That makes them look a bit better and also reduces file size by quite a lot. Also, while porting him to our new costume color system we changed the aesthetics several of his costume colors. Check out the new choices!
--Updated visual effects for his projectile. Implemented effects for his big projectile (hold B) and electrified projectiles too (B during wind). (There will be new effects during the starup and impact of these fireballs later.)
--Changed behavior of big fireball during wind. Months ago, it got powered up and was too good. Recently, it was not powered up at all, which feels wrong. Now it does get powered up during wind, but it only travels a short distance before dissipating.
--Updated visual effects for lightning strike of air B during wind.
--Grave no longer spends wind meter unless the wind really happens.
--Doing wind (forward) when near the ground no longer makes Grave land before actually summoning the wind.


--Ground super now does 2 damage rather than 1 damage.
--updated character select screen idle pose and “I’m selected” animation


--updated character select screen idle pose and “I’m selected” animation
--Fixed a bug where Rook’s air C statue wasn’t showing the correct color palette to match Rook’s costume color.
--Thunderclap now has 4 fewer active frames and 4 more recovery frames.


--New animations for almost everything: idle, f+A, A, b+A, C, jump B, and jump C. Also ground super and air super. Also, forward throw and back throw. (She still has placeholder animation for walk forward and back, jumps, knockdown and get up, and yomi counter.)
--Old placeholder paint effects replaced with new placeholder paint effects that are more paint-like.
--Hitboxes for C and air C are normal sized and not crazy anymore.


--Very slight changes to costume colors to fix some that were showing up too bright (like radiating extra light).
--Changed default costume Setsuki’s inner pants from red to dark brown.


--Fixed a bug that made flashing life persist after getting hit by an attack.
--Fixed a bug where quitting a game during a cinematic camera angle would use that same camera angle for normal gameplay the next time you returned to gameplay.
--The game’s window can no longer be resized. it’s always 16:9 at fixed resolutions now. Later we will enable resizing it to other 16:9 sizes. 
--New UI feature for menu help text implemented. On the “play” submenu, the new help text for “back” can now be mouse clicked to go back.
--You can now switch between using the UI in “mouse and keyboard mode” and “console controller mode” in the options. As a debug feature, pressing the C key also toggles this.
--Character select screens that involve only one character being selected will have a new layout. You can see a preview of this in Play -> Casual Play. In later builds, this screen will be more fleshed out and will take the place of the character select screen in Arcade mode.
--Several other small bugs fixed.


Play this build right now at the $25+ support level on our Patreon.

Overwatch's Ranking Point System

Overwatch team: great job and all. If you want to listen to an hour of me saying great job, here's a podcast about that.

You should probably re-think the current system of gaining/losing rank points though. Specifically, adjusting the ranking based on individual performance rather than just win/loss is pretty dangerous.

Elo for Team Games

Elo is a standard ranking system. You gain points for winning and lose points for losing. Furthermore, you gain more points for beating someone ranked higher and you lose more points for losing to someone ranked lower. Elo is designed for 1v1 games though, not team games.

To generalize Elo to team games, there's two factors you'd use. First, if your team's AVERAGE ranking was lower than the opposing team's average ranking, then you should get more points for winning. Second, if your PERSONAL ranking is lower than your team's average ranking, you should get more points for winning than your higher-ranked teammates who also won. As far as I know, all of this is true in Overwatch and makes sense.

But what about your individual PERFORMANCE during a game? For example, you lost but you played really well and your stupid teammates caused you to lose. Should you lose FEWER points for this loss because you personally played well? This is dangerous territory. If your instinct is to say yes, then at least consider that this requires you to gain fewer points for a win if you happened to not play well. That's really the least of our worries though.

Before we get into adjusting ranking based on individual performance, does Overwatch currently do that at all? The answer appears to be yes. Here's an excerpt from this article:

Cloud9 carry Lane “SureFour” Roberts was the first player to hit 80 Skill Rating when Competitive Play launched, queuing almost exclusively with his professional player teammates. When Roberts finished his 10 placement matches, he received a 77 Skill Rating, commensurate to his talent as one of the best players in Overwatch. Cloud9’s support Adam Eckel, who played the exact same 10 placement matches as part of a Cloud9-stacked queue, only received a 67 Skill Rating. His tank teammates hit 71. Derrick "reaver" Nowicki, the team’s other carry player, hit 74. All of those numbers rank in the top one percent of Overwatch players, according to MasterOverwatch, but that’s a pretty big discrepancy for players who contributed to winning the exact same games against the exact same opponents.


So what happened here is controlled test where a team of 6 players only ever played with each other and necessarily had the same win/loss record against the same opponents. Because they ended up with different ranks, it looks like individual performance really does matter in this equation. There could be some other explanation maybe, but it's highly likely that their individual performance metrics is what explains the difference in ranks.

Microsoft TrueSkill

Generalizing Elo into a system that handles team games isn't new. That was exactly the purpose of Microsoft's TrueSkill ranking system over a decade ago. TrueSkill intentionally and explicitly does NOT use any individual performance metrics. Their argument is that no matter what game you're talking about and no matter what metrics you measure to determine how well a given player did, it's necessarily imperfect compared to using only win/loss. The point of trying to guess if a player did well or not is how much they contributed win/loss, but the win/loss stat is the most accurate measure, they say. You'd introducing error by adding ANY other metric.

In addition to introducing error, you're warping incentives. For example, if you measure "damage done" as one metric, then it means players will attempt to maximize "winning AND damage done" rather than just "winning," which is not great. You can also very easily accidentally do a lot worse: you might accidentally give incentive not to play support heroes in a game where you really need support heroes on your team. (It seems this is already true in Overwatch.)

In many cases, it's almost hopeless to even devise a metric. If a character's role has to do with healing, you can't actually use how much they healed as a measure of much. If you did, it would penalize a healer on a team that played so well they didn't need as much healing. Or even worse is a character like Mei. Her ice walls can do a lot, her slow and freeze effects can do a lot too. But to actually quantify that into a metric correlated to win-rate? That's a huge error effect waiting to happen. My friend suggested the best metric how effective you were with her is to monitor the opponent's chat to detect how much they are cursing about Mei.

Yet another issue is that it's easy to accidentally create competition within a team for no real reason. For example, if number of kills is a metric that affects your rating, then your teammate killing an enemy that you could have killed essentially "stole" ranking points from you. That's clearly a bad dynamic.

I think Microsoft TrueSkill's reasoning makes sense here. It's a good case against ever using any individual performance metric when adjusting ranking points after a win or a loss.

Tangent: Another Thing about TrueSkill

You can skip this section if you're just here for Overwatch stuff. I just wanted to note that I'm not fully behind the REST of what TrueSkill does. The main idea behind TrueSkill is rather than assign a specific ranking number to a player, behind the scenes its assigning a bell curve probability distribution of what it thinks about your ranking. So two players might both be ranked in the 54th percentile (about tied) but the system strongly believes that's correct for player1 while for player2 it has a wide bell curve showing very low confidence in that ranking.

In theory, I see how this would allow it to converge more quickly to a good value. And in empirical tests done by Microsoft, it did converge faster than a more Elo-like system that didn't use the probability stuff. But...it just seems wrong anyway.

Specifically, if I beat a player way better than me (according to our ranks), I expect to go up a LOT of points. If I go up very few points "because the system is very sure of my current rank," that feels like total bullshit. And I have had this exact experience before. It's confusing and frustrating. As a player, I actually resent the system claiming it's so sure about me and dampening my rank gains when I go against its expectations. I think that feels debilitating and doesn't work well in cases where players really do get a lot better.

Anyway, I don't think Overwatch is doing this.

In Favor of Individual Performance Metrics Affecting Rank

Even though it sounds like a bad idea to count individual performance metrics when adjusting ranking points, is there some reason to do it anyway? Yes, I think there's something in the plus column here. The two main plusses I know of are "good outweighs the bad" and "assistance to escape Elo hell".

Good Outweighs the Bad (??)

Yeah it's imperfect to add any metric at all that gives you a bonus for kill:death ratio or whatever rather than just win/loss, but maybe it helps more than it hurts. For a character like Reaper, kill:death ratio is a relevant metric. It's not a perfect one for sure, but if you did amazing on this stat, chances are fairly likely that you played well. There are times where this indicator is wrong, but we might beat the baseline of "never count this stat" more often than we'd be steered wrong if we do count this metric for Reaper. I don't know that to even be true for real, but that's an argument someone could make.

I think the trouble here is that it's playing with fire. It's very easy to mess this up, so the downside is clear. If you mess it up, you get situations like the Cloud 9 example above where support players appear to be punished accidentally due to the workings of these behind-the-scenes algorithms. Is that risk really worth it? The upside is helping your rating converge to a good value more quickly, but maybe that's less important than avoiding these potentially very bad downsides.

Elo Hell

A related point here is about the urban myth of "Elo hell." This is the phenomenon where players with bad ranks in a team game can't rank up, even though they are actually much better than their current rank says. Their bad teammates make them lose so much that they can't rank up.

Is this a real phenomenon or just a myth? If it's real, then don't we actually place a lot of value on having individual performance metrics boost these decent players out of their unfairly low ranks? After all, they are playing great so they deserve some ranking boost.

I think Elo hell actually is real...sort of. Let's start by looking at the part of it that just can't be real though. If you are actually much better than your rank, then in a 6v6 team game you'd expect on average to have 5 "bozos" on your team, which is one less than the 6 bozos on the opposing team. So...just play enough games and your ranking will climb. Surely you are providing an advantage to your team in getting wins, because that's the very definition of what you being good means. If you find you can't get over 50% win rate, it sounds like you are actually as bad as the other bozos?

Mathematically, that makes sense. But let's look at this in the form of a story to truly understand it. You are playing as Reaper on a payload map. You decide to teleport behind enemy lines, then sneak up on various players. You're trying to flank them, catch them unaware, and get in kills. The more successful you are at this, the easier you're making it for the rest of your team. You aren't at the objective here because it's not your job. Your job is to make it so your teammates at the objective have a really easy job.

You get 3 kills in quick succession and you don't die. How are you doing? I think you're doing incredibly well. Your plan makes sense and your contribution is very large. If you had killed just one player, you might have pulled your weight at least, but by killing 3 there are now only 3 players left on the enemy team. Surely because of this, your team now has control of the payload.

You look at the payload indicator UI, expecting to see three arrows from your team pushing it forward. Instead, you see one arrow of the opposing team pushing it back. You wonder how this is even possible, so you go to the payload. What you see is a single enemy Reinhardt standing on the payload, totally unopposed, with no other players on either team in sight. Welcome to Elo hell.

You end up losing this game. The situation described is pretty unreasonable though. Your stupid teammates should have capitalized on the advantage you gave them and taken the payload. Instead they chased down butterflies or whatever and failed to get any real value out of your contribution. It's actually quite easy to imagine situations like this happening over and over such that even though you do amazing stuff, you still only in around 50-50. So in this sense, yeah Elo hell is real.

I think there's more to the story though. If we try to address this by rewarding you for your good individual performance and to get you to your "rightful" rank, we run into a couple problems. As stated earlier, if we reward you for number of kills, or K:D ratio, or damage done, we also introduce warped incentives. Now your incentive is something OTHER than just winning. Now you're fighting with teammates for kills, etc. So even if we wanted to help you out here, it's dangerous to do so.

But even beyond that, SHOULD we help you out? If we do, the result is that you are going to gain rank for doing things that...didn't help your team win? Yeah it SHOULD have helped your team win, but it didn't. It's a bit weird that you'll then keep playing the same, keep not actually making your team win (even though it's their fault), and we reward you.

Here's the real truth about this Elo hell stuff I think. The example Reaper situation above really is good play, it really is something that should help the team win...if you were a higher rank. The higher rank all the players involved, the more easily your teammates can convert advantage you provide into a win. If your teammates are so bad that they can't convert the advantage you gave into a win, then you should do some completely different things. Yeah it sucks that the thing you did SHOULD help, but in truth, it didn't. Work with what you have. Work with your generally uncoordinated or lower-skilled teammates and provide them whatever they actually DO need to win.

In Overwatch, I think what players generally need in these situations is "babysitting." What I mean is, it's probably more important to have few deaths and to generally be on the payload than it is to achieve impressive stats that "in theory" allow your teammates to be on the payload.  You have to carry them, so you'll have to refrain from strategies that, at higher rank, are very good, so that you can provide for the most basic needs of your team. You don't have to do that in the exact way I said, but the point is if you play in the (sometimes pathetic) way that your team needs, you can contribute more to your team's win rate than if you play in an incredibly impressive way that they are unable to capitalize on, because they suck. Yeah that's frustrating, but THAT is the way out of Elo hell. Having the system give you a ranking boost for strategies that aren't resulting in a positive win rate isn't a good solution.

I don't now the reason Blizzard chose to have individual stats count towards rank (or even 100% that they do, but they sure seem to). I'd advise against it though.


Sirlin on Game Design, Ep 16: Overwatch

We discuss Overwatch, Blizzard's first-person shooter. It far exceeded our expectations and we attempt to explain how it manages to do that. This isn't a "review," but rather us analyzing the design decisions involved, trying to define the secret sauce of Overwatch's success.

Hosts: David Sirlin and Matt "Aphotix" DeMasi