The Bazaar: Bad Metagame Design

The Bazaar (currently in closed beta) is a Skinner-box shopping autobattler, in which the player buys and sells items from various ‘merchants’ before having their board face off against an asynchronous snapshot of another player’s board at the same point in their ‘run’. I’ve been enjoying it quite a bit.

The game looks promising, but its reward structures and metagame design are shockingly hostile to new players.

During a run, the player goes through a regular cycle of merchants and NPC battles before facing the recorded ‘ghost’ of another player at the end of each ‘day.’ A victory fills in one of the little spaces on your victory tracker, with the 10th win ending the run in a ‘Gold Victory.’ A loss lowers your prestige, which begins at 20, by a number equal to whatever day it is: lose on Day 3, lose 3 prestige. The first time your prestige hits zero, the player is given a choice of bonuses to help them pull their shit together. With the next loss, the run ends.

‘Ranked’ mode requires a ticket, which can be earned as a reward in Normal mode, pieced together from gems only obtainable as rewards from Ranked mode, or (naturally) can be purchased for USD $1. Players also get a single free entry to Ranked mode per day. Runs in Ranked mode earn one ‘chest’ for the 4th win, a second for the 7th, and a third for a Gold Victory at 10. Purchaseable battle passes will double these rewards.

‘Normal’ mode has no cost of entry, and rewards the player with a single ticket to Ranked mode for a victorious run of 10 wins. No chests, no gems, no ticket ‘fragments,’ no other rewards of any kind.

While I respect Mr. Yanyuk’s desire to make staggering amounts of money from battle passes, pay-to-win expansions, and NFT cosmetics, this reward structure exerts some very troubling behavioural pressures on the player.

If we assume that the matchmaking is fair, and disregard ties (both boards dying on the same 0.2s ‘tick’ counts as a victory for the player, as does both boards surviving the round-timer sandstorm), then the expected 50% winrate means that the median player’s run will end at Day 10, having won a mere 5 battles. New players are, naturally, not expected to perform as well as the median player.

What this means is that most new players will not have a snowball’s chance in hell of making it to 10 wins with any kind of regularity. Instead, they are repeatedly kicked down and told they’re not good enough to deserve anything for their pathetic 4 wins, or even their personal best of 8 or 9. And 9 wins isn’t necessarily all that close to victory: due to how health and damage scales from day 10 onwards, that 10th win can prove a great challenge even to those a bit more practice under their belt. Many players have voiced frustrations already over the repeated experience of sailing undefeated through the first nine days, only to face the juggernauts of Day 10+ and have their run end with three back-to-back defeats. And then be told their efforts weren’t worthy of any reward at all, and that if they can’t hit 10, they might as well not even try.

You can start to see the problems here.

In Ranked, players are incentivised to favour consistency and play struggling runs through to the end.

In Ranked play, each run has an associated cost (the highly-limited ‘ticket’ or $1 entry fee), which would be lost by restarting.

Additional rewards at 4 and 7 wins provide interim goals and an incentive to continue and make the best out of bad situations.

The ‘last stand’ reward is often of low utility, but could be an immense boost that suddenly makes 10 wins likely again, incentivising the player to play until the last stand just in case. Then, even if that last stand reward is useless, conceding now saves so little time that the player is incentivised to just play it out and hope for a miracle.

The rewards at each threshold—4, 7, and 10 wins—is exactly the same. Given the difficulty of achieving those last few wins, this incentivises lower-risk strategies that usually walk away with 7 wins, over high-risk strategies that may get 10 occasionally, but usually fall short of the second chest at 7.

In Normal mode, players are incentivised to favour high-risk strategies and restart until they get lucky.

There’s no cost associated with restarting a run, so players are incentivised to just bail on a run that’s not going well instead of trying to see it through.

Without interim goals, a rocky start leaves a 10-win run looking insurmountable, incentivising the player to just bail and start over. (NB: reduced prestige loss in early days strongly mitigates this, but the effect is lost once the player becomes familiar with how cracked enemy builds become after Day 10)

The strength of ghosts increases rapidly after Day 10. Multiple factors contribute to this:

• Snowball builds that start weak and get stronger over time have by day 10 well and truly crossed the threshold of being ‘good’ and are starting to enter the stratosphere

• High-risk strategies that have paid off are retained in the population, while high-risk strategies that failed have been excluded by selection. This effect of player elimination dramatically increases the representation of ‘cracked’ or ‘lucky’ boards in later days, even when those strategies have comparable or lower win-rates.

• Player concessions prevent their weaker-than-average ghost from being available in the future days they would have reached, but didn’t play (i.e. the last few losses before the game formally hands out a defeat)

• Prestige maths means that weak/unlucky runs bomb out very consistently between Day 8 and Day 10. In fact, even a run of 100% losses still makes it to Day 7, at which point a recovery would require that the player not only streak the next 10 wins undefeated, they would need to survive until Day 17 while doing so… an utterly ludicrous proposition. This means that even if all players always saw their doomed runs through to the end, a huge proportion of weaker ghosts disappear from the population quite suddenly in the lead up to Day 10.

This means that not only is losing a run incredibly slow, but every loss makes recovery much harder. It’s no wonder that conceding is rampant—it’s a lot easier for players to play the first day over and over until they get something they’re happy with, rather than trying to forge ahead with a bad (or even mediocre) start. In Ranked mode, the rewards for four or seven wins still seem achievable after a few losses, but in Normal mode the complete lack of any reward for anything less than a perfect 10 tells the player pretty explicitly, “you should just quit. Even a Herculanean effort here will go completely unrewarded.”

Why is this a problem?

Any competitive zero-sum game requires a healthy population of new and unskilled players, to prevent attrition of the least successful players inevitably leading to a tightly bracketed playerbase of high experience and skill, creating a game experience completely impenetrable to new recruits. This can, to a degree, be mitigated by the self-segregation of the most skilled players into Ranked mode, but even then it should be noted that The Bazaar currently has no rank-based matchmaking, and no plans to do so.

These new players, naturally, head to Normal mode first to learn the ropes—it has no cost to enter, and the very concept promises a more forgiving atmosphere, free from both the most experienced and the most competitive players. A more fun, more relaxed, and lower-difficulty environment in which to learn the game and gain confidence before trying their hand at Ranked.

Instead, what they find is a brutally punishing environment where even their best and luckiest runs get absolutely crushed by ridiculously cracked boards they have no hope of learning from or replicating. And then the game gives them absolutely nothing for their efforts, and says that if they can’t hit the completely unattainable goal of 10 wins, they shouldn’t even bother trying.

If, somehow undiscouraged by this, they forge ahead, learn the game, practice until they’re comfortable in Normal mode, and decide it’s time to ‘graduate’ and try their hand at a competitive Ranked experience, everything they’ve just learned in Normal mode has now set them up to have a terrible experience. The cost of entry prevents them from bailing like they want to, and all their learned strategies and risk-profile behaviours—which events to choose, how much to reroll shops etc—are actively harmful.

The final word:

The Bazaar has the makings of a great game, but unless the designers recognise the criticality of these issues and make some major changes, it won’t survive for long. Aligning player incentives between the game’s casual and ranked modes is essential, and unless this issue is addressed, the game’s new-player attrition rate come release is going to be horrific.