22nd May, 2014By Softmints5 minutes

I'm sure most of you are familiar with the concept of drafting, the pre-match ritual in which players negotiate the characters that will be in-play during a match. Typical drafting phases consist of an alternating sequence of picks and bans. I've listed the drafting sequence for several commercial games below. There's some variation, but they're all pretty similar:

A grid of icons showing the pick/ban sequences for several popular mobas

These complicated sequences with picks and bans are usually considered the most balanced mode available, and as such is used for tournaments or matches with pre-arranged teams.

I'd like to point out some trivially observed "mechanical" properties of the format:

  • Bans are offered before the first pick. Teams can choose to avoid playing against certain heroes, if they deem them too strong in the meta, or want to limit their opponents' access to counter-picks. This also serves as an automatic balance-correction mechanism to keep the game playable between patches.
  • Initiative changes frequently. Each team has the opportunity to see an increasing portion of their opponent's team composition before increasing their commitment to their own strategy. As a result, it's difficult to run all-in strategies (such as a zergling rush), since the opponent can see it coming and has a chance to prepare. In this way, the formalities of drafting reduce the potential impact of drafting.
  • Picks are unique. Once a character has been picked, it cannot be picked again. This encourages each match to have a diverse selection of heroes, and that each team will have an unequal set of advantages and disadvantages on which to capitalise. The resulting gameplay is more dynamic.

We can also make some observations about the impact drafting has on the game:

  • Personalises the game. Using bans, teams can steer the picks in a direction that favours their preferred strategy or style of play, which allows different players and personalities to flourish.
  • Keeps the meta fresh. Without dynamically removing some options, the meta might devolve too quickly.
  • Rewards valuation. The drafting phase is celebrated because it allows players to demonstrate and be rewarded for their deep knowledge of the game at a macro level. The timings for certain picks and bans is also a skill in itself.
  • Builds tension. Not to be underestimated, negotiations under time-pressure can build anticipation for the game to come.

So there are some great advantages to drafting in general. However, there is one aspect which has some downsides: the bans. It doesn't feel good to have the hero you wanted to play taken from you.

The only way to avoid getting caught is to play a mode that doesn't have bans, but such modes are less balanced, and aren't an option in a competitive setting. Bans also have some subtler negative repercussions:

  • Competitions should be an opportunity for players to demonstrate their skills when they're at their best. Bans produce the exact opposite: if you get good at something, you don't get to play it. It might free up other options for your team, but it doesn't produce or reward the best play.
  • Similarly, if a team comes up with an innovative strategy and proves that it's effective, their opponents can thoughtlessly and immediately respond by banning the heroes that were used and force the game back into a more familiar meta. At best, some usually banned heroes can be picked, and each team will get an equal shot at picking those. The strategist doesn't see much reward for their innovation.

Most players, and particularly competitive players have come to accept that learning and playing multiple heroes is unavoidable. It's easily rationalised by claiming that any skilled player would need to be diverse to avoid counter-picks anyway.

Which is generally true, but something that we'd rather see emerge, than be enforced. Is there any way we could remove bans, without taking away from the things which make drafting good?

To address the above fully, we should also be removing unique picks; allowing both teams to have the same hero. This can be done, but risks producing degenerate gameplay (both sides pick mostly the same heroes and have less advantages to exploit).

There's also the risk of conflicting audio/visual cues since the same hero appears twice (unless, as in Bloodline Champions or Prime World, there are two copies of every hero).

I'd like to propose a new drafting mode to address bans. You've seen All pick, you've seen Captain's pick. Introducing: 

"Cherry Pick" in a script font, with cherry foliage in the background

Cherry Pick is a drafting mode which has no bans, and optionally supports mirrored picks on each team. Its premise is founded on the negotiation element of regular drafting: that part is a definite keeper. But if by assumption our opponents can always pick any heroes they want, how can we possibly exhibit control over their choices?

Simple. We influence the order in which they make picks. It's common in drafting that we don't want to give away our strategy too easily: we try to conceal information for as long as possible to avoid being counter-picked. How about placing that information war centre-stage?

Rather than bans, I'm introducing a new step called an offer. An offer takes place in three steps:

  • Firstly, one team highlights a hero, and offers it to the enemy team.
  • The enemy team can either pick that hero immediately, or decline the offer.
  • No matter what they chose, the highlighted hero is then removed from the hero pool and can no longer be picked by either team.

This creates a "now or never" situation, where if a team really wants a certain hero, their opponents can force them to take it early in the draft, where it will reveal more information about their overall strategy. The sample sequence below can terminate in 10 steps, or if every offer is declined, takes 20 (same as the current Captain's Pick in DotA).

A sequence of icons, with a 'present' icon taking the place of bans

When a team accepts an offer, they skip their next 'pick' to ensure that the initiative keeps changing. The first set of 12 steps will ensure each team has 3 heroes, before progressing into the next set of 8. There is plenty of room to adjust the specifics.

What can we say about Cherry Pick?

  • If a team wants to pick a hero, they are guaranteed to have a chance to get it, simply by not offering it away. This also makes the mode serviceable for casual play, since it accommodates players who want a specific hero.
  • Without enforcing unique picks, we tend to end up with picks that are unique anyway due to offers. This is great for diversity.
  • If there exists an overpowered hero in the pool, it's unlikely to be offered away, and hence both teams will be able to pick it up during regular picks. This is also an automatic balance-correction mechanism, though games will tend towards always including overpowered heroes, rather than never.
    • A possible amendment is to have an initial step in which both teams blindly nominate up to 5 heroes, and any heroes nominated by both parties are ruled out. This helps players self-regulate to keep overpowered heroes out of their matches, and since it's opt-in, doesn't interfere with our goals.
    • Without the protective layer of bans, the game's balance is under direct scrutiny, and must be managed properly.
  • There's still a good negotiation process, keeping many of the positive features of regular drafting. The system builds on the existing complexity of counter-picks.
  • If we know for certain that the enemy team will pick a specific hero, we can target our offers on the less certain "supporting cast" and still gain valuable information about the enemy draft.
  • Even if every offer is refused because teams are terrified of commitment, we still have a fairly dynamic drafting phase (the offers were functionally "bans").

This mode hasn't been subject to competitive scrutiny, so perhaps there are holes. It might be optimal to offer easily-countered heroes first so your opponents never accept, or it might depend entirely on the particular moba and its heroes.

Adjusting the sequence of offers/picks might be one way to take weight off the earliest offers if there's a problem.

The name is derived from the nature of the mode: of the offers available, you only accept the ones that you really want (enough to commit to them early in the draft).

I don't mean to give the impression that bans are the devil; in practice drafting with bans is still a workable system. It is because they are so accepted, that when the idea of removing bans from drafting came to my attention, I really wanted to try tackling it. 

I'm pleased with Cherry Pick, and would definitely like to iterate on it further.

This article was motivated by a discussion among the FantasyStrike.com community, which laid out most of the points in the first half of this article.

Further reading: Sirlin on Game Design: Character Bans