Last time I mentioned that Ram Wrangler was one of my least favourite cards. The reason that I gave at the time is I didn't like the amount of variance in the card. It can summon anything from a 1 mana to a 9 mana card directly onto the battlefield. To me, as a designer this feels not very good, but I couldn't put my finger on exactly why that was until I watched this new Extra Credits video:
Go watch it. I'll wait...
...okay, back? The video introduces a concept called the Delta of Randomness. First, it posits that a good random card at its worst should be a below average card, and at its best should be an above average card. It goes on to say:
"[delta of randomness] actually refers to a given card's maximum and minimum distance from the power curve."
The more above the curve the best result can be, the worse its worst result must be, otherwise the card is just too good. So far so good, right? Here's where things go bad:
The larger the gap between best and worst result the more likely that result will determine the fate of the game. As the video notes:
"The problem is often these moments [of extreme variance] are exciting for only one player. A bad result on the randomness table, in this situation, is fun for nobody, and a good result may give one of the players a bit of a rush, but it's just going to give the other player the feeling like game is unfair (and that's not the result you want as a designer)."
The video summarizes its thesis by saying:
"For every game there is some line past which too high a delta creates a feeling of being arbitrary rather than exciting."
So let's get back to our Ram Wrangler. There are 48 Beasts as of LoE in the game, which vary from 9 to 1 mana, from '1/1' to '8/8 Charge.' The Beasts are put into the battlefield, obviating the need to spend mana on them. We pay 5 mana for anywhere between 4/4 and 11/11 worth of stats. The delta of randomness in this case is extreme, and the card is bad for the game. As the video points out:
"When the cards aren't quite as swingy, players don't immediately dismiss the matches result because of one good or bad roll of the dice.
This points us in the direction of how we, as designers can do better. The new Discover mechanic is one such way that Blizzard is permitting randomness, but reducing the delta. With 3 draws, instead of 1, Team 5 is making the quality of the randomness more consistent, and allowing players to contextually choose the best card from a list, rather than enforce an absolute variance from worst result to the best. Discover also frequently reduces the overall pool of possible choices, making a certain quality of result more predictable.
The video also a suggests a sweet spot with their 20/60/20 guideline. In short, when an instance of randomness is invoked 20% of the time the result of the RNG shouldn't really matter, 60% of the time it should provide an incremental gain, and the final 20% should have both players on their feet, since the outcome of the game may hinge on the result of the die roll. While I don't believe the exact percentages should be stressed over, I think this is a good model to prompt questions about the RNG we introduce into Hearthstone when designing:
- What is the delta of randomness of our card?
- Is our worst result below the power curve?
- Is our best result above the power curve?
- Are the odds of worst and best result equally above and below the curve?
- Are there instances where the RNG is
- Meaningless?
- Incrementally meaningful?
- Possibly game-deciding?
- Are the ratios of meaningless, incremental and game-deciding roughly in that 20/60/20 mould?
In light of this video, I thought I'd go back to one of my previous designs with a random element --Warsong Tactician:
- The delta of randomness varies from having no target or a minion that already has Taunt, therefore no effect, to targeting a minion that's got high Health and is difficult to remove. The dream is getting this when your opponent is one off lethal, and it targets a minion with Taunt that can clear there entire enemy board when they have no other answer.
- A 4/4 for 4 is below average for class card, which could be a 4/5 with a very small effect.
- A 4/4 that gives a good minion taunt when drawn is certainly above average, since the cost of giving Taunt is effectively 0.
- It's very hard to assess when this Taunt-giving effect is going to game-deciding, but I'd imagine that around 40-50% of the time you'll have no board when this card is drawn. This puts it on the poor side of the 20/60/20 scale (more like a 40/50/10).
Conclusions: While I once strongly believe that it was balanced at 4 mana 4/4. The delta of randomness seems to fall within acceptable, and possibly punitively underpowered bounds. I'm now more inclined to make it 5/4 with the same effect.
I highly recommend subscribing to Extra Credits and going through their sizeable backlog of videos. There's a lot that group can teach us about good design. If, however, you're only interested in the Hearthstone stuff, there are two other videos well worth watching:
As always, I love having conversations, so head on over to the Reddit post and let's talk!
No comments:
Post a Comment