Steam reviews: issues & requests

FindingThe complaint is not that the game is random but that there are too few affordable levers to correct the randomness: dice placement and the shop pool are both random while the currency that would let you steer either is capped and contested. On a bad run players conclude their decisions stopped mattering to the outcome.

• 55% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• What recurs is the shortage of mitigation, not the presence of luck — the resource for cycling the shop is “CAPPED - and also competes with the need to use it for dice assignment or triggering artifacts,” so the one corrective lever is also what pays for everything else.
• Randomness stacks in two layers, the shop and the dice themselves: “the primary way of interacting with the board is also rng so thats another source of not interacting with the game.” Players add that placement is pure random and the count is fixed at three, with no pip-changing or extra-dice option that comparable dice roguelikes offer.
• The recommend split tracks whether a player ever accumulated enough manipulation, not whether they minded luck: recommenders describe earning it — “you earn the rng manipulation to receive your shop items through skipping or other means” — while detractors report the levers never arriving.
• Pacing compounds the scarcity: “The steep score ramp up means the early game goes by so quick you don’t really get a chance to establish a strategy for the run,” so early investment is written off and pivots follow whatever the shop offered rather than a plan.
• The five Chinese-language sentences name mechanics rather than feel and concentrate on the highest difficulties — an oversized relic pool, key pieces and relic locks unreachable, three fixed dice with no face manipulation. All four schinese come from non-recommenders, one asking what is left to play in “a luck-dependent game that also deliberately destroys the player’s build” (translated).
• The sharpest statement of the shared objection is about outcome, not variance: “I can’t quite shake the feeling that my decisions are not contributing to my success (or failure),” which is why even several recommenders raise it while still endorsing the game.

The 20 mentions are not 20 complaints — 11 come from recommenders, several of whom raise RNG specifically to argue it is manageable, so the theme’s size overstates the dissent.

FindingPlayers are not saying the presentation is bad — they say it is scaled back from StarVaders, with a short music loop and animations that play out identically regardless of commander, pilot or outcome. The art style itself is repeatedly praised; what is attacked is the lack of variation inside it.

• 89% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The frame is comparative, not absolute: the same reviews that complain call the art “full of character” and the presentation “superb,” and seven across both languages measure it against the prequel. One states it most precisely: “Starvaders' presentation, soundtrack, and extra fluff was so outstanding that it's something I miss here.”
• Nobody calls the music bad — it is “chill,” “catchy,” “banging,” “immaculate.” The recurring ask, in six reviews, is for more of it: “wish there was a variety of options because the loop has been carved in my brain at this point,” echoed in Chinese as 「BGM比较少」.
• The animation complaint is specifically about sameness, not quality: they “aren't customized to yoru character or the enemy pilot really at all,” and two more reviews ask for per-commander and per-pilot animations by name.
• The five Chinese reviews ask for set-piece moments rather than more variety sliders — a defeat CG (「可惜的是游戏没有战败CG」), more 演出 (staged presentation), and a victory animation worth watching, since the current one is “基本上每次大差不差的最后都跳过了” — near-identical every time, so it gets skipped. That register does not appear in the English reviews.
• The “one screen” half of the theme title is carried by a single review, which asks for backgrounds that shift as a run progresses rather than for more screens; no one else raises it, so it is a wish, not a shared complaint.
• Presentation is not what drives the negatives: both Chinese non-recommends explicitly credit the art and sound (「美术音效足够出色」) and place the fault on mechanics, and the one English non-recommend raises animations as a secondary gripe.

Of the 23 sentences, roughly five are pure praise of the art or OST caught by the same keywords, so 19 mentions overstates how many people are actually complaining.

FindingPlayers are not saying the top difficulty is too hard — they are saying its hardness is made of restrictions (locked artifact slots, stripped feats, bosses designed to invalidate whatever you assembled) rather than new mechanics to play against.

• 59% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The objection is to the method, not the level: difficulty 6 answers the optimal-build problem with locked artifact slots, stripped feats and stacked challenges, and the recurring reply is that this is not content — “if it's just imposing limits, players can do that to themselves.”
• These restrictions land on an already luck-dependent core, which is what turns irritation into a refund-shaped complaint; one reviewer states the shared point most precisely: “a game this dependent on luck, and it deliberately breaks the player's build — so what exactly do you want players to play?”
• What players expected from top-difficulty bosses is new mechanics, and what recurs instead is anti-build bosses plus reward rules that punish playing well — a four-turn boss that withholds its reward if you clear too fast, “no reward for completing it, a penalty for not.”
• Blocking builds narrows the game rather than widening it: multiple reviewers report falling back on the same handful of demo-era builds, and the one posted difficulty-6 clear guide is a single raw base-attack stack chosen precisely because it needs no feats, no shop rarity and no synergy.
• English reviews report a materially different problem — Act 6 as a tuning spike and flat scaling (“Overseer straight up doubling all score thresholds”), with one reviewer wanting the peak to be higher still — while the build-blocking reading is carried by the 11 Chinese reviews and the single Japanese one (“the high-difficulty extra rules aren't interesting”).
• The constellation system draws the same objection in a different wrapper: two reviewers say a mechanic meant to add relics and variety instead puts the upgrades a build needs out of early reach, so “you just die without ever seeing the piece.”

17 mentions is a small theme confined to optional late content, and 58.8% still recommend — several of the sharpest build-blocking complaints come from positive reviews that amount to “good game, don't touch high difficulty.”

FindingPlayers are not calling the runs repetitive so much as the optimal path repetitive: build choice collapses into one dominant line once difficulty rises, and nothing new arrives after the endgame. Nine of fourteen still recommend, so this reads as a ceiling complaint from engaged players rather than a reason to quit.

• 64% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The early and mid run are largely not the problem — recommenders say the opposite, that “the game heavily incentivizes pivoting your build prevents individual runs from feeling too boring or brain-dead.” What recurs is that this freedom closes down later, not that it was never there.
• The sharpest recurring claim is build convergence: “这个游戏的最佳路径收敛得太快了” (the optimal path converges too fast), echoed in English as “the core loop of peashooter, gluttony, score is so consistent it makes the late game pretty bland.”
• Chinese reviews uniquely tie that convergence to difficulty scaling — at difficulty 6 the numbers fall behind enemy growth unless you play the one line, leaving only “反复刷新等核心棋子然后不断重复回合” (repeatedly rerolling for core pieces and repeating turns). English reviews describe the same monotony but never name a difficulty, so the forced-optimal framing is schinese-specific.
• The second strand is that nothing new follows the endgame: “唯一问题是终局之后缺乏有意思的玩法” (the only problem is the lack of interesting play after the endgame) and “the late game gets a little monotonous” — both from players who still recommend.
• New characters are the specific disappointment in the 1.0 content: numerous but thin, “新人物虽多，但都没什么意思，就体验他的被动？” (many new characters, but none are interesting — you just experience their passive?), and one English review frames the release as the demo “played with 6 other characters.”
• Counter-evidence is real and recurs too — multiple reviews praise content volume, the per-character strategy space, and in Korean that the player scales as fast as the numbers so it “지루하지 않다” (isn't boring); the complaint is about the top end, not the whole game.

Of 14 mentions only 5 did not recommend, and several of the repetition sentences come from positive reviews describing a late-game ceiling.

FindingThe complaint is a stable power-tier gap between commanders, not that they feel alike: a few are described as obviously dominant while the weakest ones are built around dice conditions the game's limited dice manipulation and random relic pool rarely let you meet. Players want the bottom tier buffed, not the roster redesigned.

• 75% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The recurring point across English, Chinese and French is tier gap, not sameness — “角色看上去各有特点但实际上强度相差很大” (characters look distinct but differ hugely in strength) and “some characters are miles above others in terms of strength”. The shape reported is identical everywhere: a top tier and, as one puts it, ones that are “路边一条” (roadside).
• Correction to the previous Finding: the named commanders do not simply differ by reviewer. English and French still name different culprits (Hex Prince/Blood Beast), but both Chinese reviewers who name anyone name the same pair — “团块和女爵简直是断档的强” (Blob and Baroness are a tier apart) and the new “建议团块加女公爵，纯粹的轮椅，一路平推” (take Blob plus the Duchess — pure wheelchair, steamroll the whole way).
• Where players name a mechanism it is payout, not design: Japanese says “初期キャラ以外は出目の要求が多く、ダイス操作は限られているためほぼ活かせません” (non-starter commanders demand specific rolls that limited dice manipulation can't deliver), and Chinese says the triple-roll identity is now “高风险低收益” (high risk, low reward). The same Japanese reviewer frames that as identity collapsing into a handicap — “実際はほぼ縛りです” (effectively restrictions) — still one voice, and the Japanese count has not moved.
• “Too similar” is the minority half of the theme and mostly points at late-game builds, not commanders: one player's “commanders are just a tad samey” sits next to another's endgame converging on “meteor spam”. The theme title weights the two halves equally; the evidence does not.
• Relic RNG compounds the gap rather than causing it — weak commanders lean on key relics from a “巨大的随机池” (huge random pool) that rarely arrive, and unlocked relics “rarely seem to show up on future runs”. That is why the stated fix is additive: “希望后续能给弱势角色相应的补强” (buff the weaker characters).
• The tone is graded, not hostile: 75% still recommend, complainants call it “My only complaint (and it's minor)” and “My small gripe”, and a returning player rates it “更平衡一些” (somewhat more balanced) than the predecessor. This reads as a tuning request from engaged players, not a verdict on the system.
• Volume is the only thing that grew: one added Chinese mention lifts the schinese share to 25%, but it repeats the existing Chinese reading rather than adding a new one.

Of the 15 keyword sentences only about eight are actually complaints — the most-helpful one praises how balancing a pivot “flows really well”, and others concern build variety or lore — so 12 mentions overstates the theme's real weight.

FindingPlayers are not complaining that a few artifacts are too strong; they are complaining that too much of the pool is dead weight, so a draft of three rarely contains a real decision and the item a build needs rarely appears. The power outliers matter only as the other half of the same problem — the choice gets made by the pool, not the player.

• 63% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The recurring objection is the floor of the pool, not the ceiling: when most offers are filler, three options collapse into one, as one player puts it — “most items and relics are horrible so when you get the one and only relic/item that isn't awful, its not really much of a choice between all of the items.”
• “Weak” usually means narrow or beaten by the default rather than literally broken — “a lot of relics feel a bit underpowered or extremely build-specific; it feels like just taking the extra income is the best play.”
• A second strand is dilution rather than power: unlocked relics “rarely seem to show up on future runs,” and a Chinese reviewer asks for a dedicated shop slot because spell, unit and relic pools share one refresh, so you roll forever for the piece or relic a run needs. A recommending player frames the same shortage as a “small pool of artifacts” and units, “especially at the maximum rarity.”
• Rerolls and locks are the designed escape hatch, and players say it stops working as difficulty rises: “Early difficulties you can reroll the perks but later on you can get stuck with a choice between 3 terrible or useless options,” with one reviewer reporting that even getting a relic lock open is a struggle.
• The rare strong artifact draws the same objection from the opposite side — a win it hands over feels “so hollow because it wasnt like it was the result of effort on my part, just choosing anything that wasn't unpickable.”
• All three Chinese sentences come from non-recommending reviews at the highest difficulty and are the only ones proposing structural fixes — a fixed shop slot, scrapping the perk reroll lock — while the English mentions sit inside positive reviews as asides (“Only complaint so far,” “and it's minor”).

At 8 mentions and 62.5% still recommending, this is a minority theme, and the single most-helpful review (28 helpful) argues the opposite — that skipping earns you enough RNG manipulation to make the artifact and unit pools feel balanced.

FindingPlayers are not objecting to exponential score targets as such — they expect those — but to boss modifiers that double the threshold on top of an already fast curve, and to an early game that ends before a build exists to meet it. The complaint is about being given no means to answer the ramp, not about the ramp.

• 50% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The sharpest recurring specific is not the baseline curve but boss modifiers that double the quota outright: three reviews across both languages name it, one as “Overseer straight up doubling all score thresholds,” another as “can't spawn bombs → target score doubles directly.”
• English reviewers tie steepness to build time rather than difficulty — the early game is over before a strategy forms, forcing an RNG-driven pivot: “it takes Balatro 30 blinds to get into the millions but you'll be there in DiceVaders before you know it.”
• When players fall short they describe orders of magnitude, not a near miss — “after playing out every round I looked at the score, then at a number several zeros bigger” — which is why the loss reads as unanswerable rather than hard.
• Chinese reviews make an argument the English ones do not, and all four touch it: late scaling substitutes numbers for design — “later difficulty has no mechanics at all, just pure number-stacking to annoy you” — and at difficulty 6 reduces play to refreshing for core pieces and repeating turns.
• This is not a blanket “too hard” complaint: half still recommend, and two of the complainers do, one describing the endless-mode exponentials matter-of-factly as something you keep up with. The objection consistently pairs with reroll scarcity, i.e. the lack of tools to respond.

Only 8 mentions and 9 sentences, two of them from the same reviewer — and the most-helpful quote here is not a complaint but a description of clearing quota early as the route to shop rarity, so treat the theme as thin and partly keyword-caught.

FindingThe complaint is not that the shop is random but that rerolling fails to resolve the randomness: the reroll resource is capped and competes with other spends, and the pool it refreshes is diluted enough that repeated rerolls still miss the one piece or relic a build needs. Players are describing a build they can already see and cannot execute, not a difficulty they cannot beat.

• 43% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• Framed as an economy problem before a luck problem: the currency is finite and contested — “the economic resource you're given to cycle through the RNG to find synergies is CAPPED - and also competes with the need to use it for dice assignment or triggering artifacts” — with a Chinese reviewer putting the same point as in-run resources where restriction exceeds acquisition.
• Spending the rerolls does not fix it, because the offers are padded: one player rerolls the shop four or five times and still never sees the piece they need, another finds the shop full of filler, and later rounds offer “a choice between 3 terrible or useless options.”
• The failure concentrates where rerolls thin out or are absent — boss relics, where “not being able to reroll the boss relics means you often times just miss on all 3 options,” and the highest difficulty, where one reviewer says you simply die without ever seeing the piece your constellation build needs.
• What recurs across both languages is blocked intent rather than lost runs: the 'On Death' build you cannot roll the items for, the relic you can never afford to lock, the constellation piece that never appears.
• Reroll scarcity also distorts the early game — the steep score ramp pushes an early pivot, and with nothing to spend “you pivot mostly based on what RNG gave you, making the early game investments feel like a waste.”
• Chinese reviews diagnose shop structure and propose fixes (a dedicated slot that only rolls movement or time points, deleting the feat reroll lock on an already-deep pool); English reviews describe the identical shortfall as a currency cap. Same complaint, different point of blame.

Only seven mentions, four of the twelve sentences come from one Chinese reviewer, and three of seven reviewers recommend the game while raising this — including the most-helpful one.

FindingThe objection is not that artifact slots are locked but that the unlock condition is a roll players cannot reach or afford: a triple is a ~2% natural outcome on three base dice, and forcing one burns the scarce reposition items needed elsewhere. Missing it forfeits the artifact reward outright rather than delaying it.

• 50% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The complaint starts from arithmetic, not taste: one Chinese review states it most precisely — "Only three base dice means the natural probability of rolling a triple is only 2%" — so the slot reads as gated behind an outcome the player cannot plan toward.
• What recurs alongside the odds is the price of forcing the roll: reposition reach is one square and 1 and 6 cannot be swapped, so "assembling one triple hugely consumes your stock of move items," which the same player ties to lost artifact upgrades and wasted turn economy.
• Failure is absolute rather than deferred, which is what turns an annoyance into a grievance: the Japanese review reports stage 1 locking the power slot behind harsh conditions where "if you don't unlock at least one in that stage, no power reward," and a Chinese player finished act 2 with slots still unopened.
• Both negative-language reviews read the locks as a designer's difficulty lever rather than a player-facing choice — the game's answer to best-path solving is "locking artifact slots, breaking up specializations, various challenges," and boss effects that "directly ban 2 artifact slots" land the same way.
• The split is by language, not volume: all three English mentions recommend the game and none object to the unlock condition, one conceding only that "the triples can feel annoying" before praising artifact exchanging. Every complaint here comes from the two Chinese and one Japanese review.

Only 3 of the 6 mentions are complaints — the two most-helpful sentences are strategy description that the keyword caught, so the 50% recommend rate overstates acceptance of the lock rather than measuring it.

FindingPlayers aren't complaining that scoring is hard — they're complaining they can't attribute the number to anything they did. The dice resolve, the total moves, and nothing on screen connects the two, so runs can't be planned.

• 50% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The recurring core is traceability at the moment of resolution, not rules complexity: one player puts it most precisely — "丢完骰子就看着分莫名其妙的上涨，或者莫名其妙的不涨" (after the roll you just watch the score inexplicably go up, or inexplicably not go up). No reviewer says the systems are too deep; they say the cause of the number is invisible.
• Two separate English reviews name the same display gap rather than a design gap — "The multiplier also isn't displayed clearly enough" and "I often have to hover over relics and units to check what they do and how close they are to leveling up." The information exists but is one hover away from where the decision is made.
• Three of the six mentions measure the game against Balatro or Luck Be A Landlord, and what they name as lost is forecasting: "significantly harder to figure out how much you will score... somewhat diminishes the strategy." The benchmark being applied is pre-commit predictability, not post-hoc explanation.
• Both Chinese reviews frame this as 莫名其妙 (inexplicable) and neither mentions UI or display at all; one extends it past the score readout to the constellation system, which "makes required build upgrades impossible to find early." At 2 of 2, that is a reading of the whole progression as arbitrary, not a readout problem — worth watching as a distinct complaint.
• This is not unanimous and should not be written up as such: a recommending review states the opposite outright, that the genre's usual "unclear interactions, scaling issues, limited viable builds" are "all seem avoided here." Half the mentions still recommend the game.

The 6 is soft — one mention praises clarity and one concerns the difficulty curve rather than scoring, leaving roughly four genuine complaints, too few to treat any split as stable.

FindingThe Balatro comparison is mostly the expectation players arrive with, not the verdict they leave; what the negative reviews actually attack is that the familiar loop is never made interesting — generic, uninspired, boring — and they say so within the first few hours.

• 40% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• Derivativeness is the frame, not the charge: three of the five reviews reach for the Balatro or “cardslop” comparison, and two of them use it to say the game cleared the bar — “I was expecting 'yet another Balatro clone' but what I got was a unique spin on the formula.”
• The complaint under the label is flatness, not imitation — “Gameplay was fun but kinda generic”, “A very uninspired roguelite with little going for it” — and one reviewer separates the two explicitly: “It's not only copying every other one out there, but has the audacity to make it boring.”
• What converts a skeptic is presentation rather than mechanics: the longest-played reviewer (31h) concedes “This is functionally the same as cardslop in a mechanical sense, but presented using an aesthetic that appeals to me, which is tiles and dice.”
• Playtime splits the theme cleanly — the two refunds-in-spirit sit at 1h and 6h, the defences at 6h and 31h — so the derivative impression forms early and only survives if the dice-and-tiles presentation lands.
• The single Chinese review carries a different complaint under the same label: “the common ailment of Balatro-likes — the higher the difficulty, the more it torments the player”, i.e. inherited difficulty scaling, not sameness. One mention, so a strand to watch rather than a pattern.

The 5 mentions come from 5 reviews across 7 sentences (two reviews contribute two sentences each), and 2 of the 5 are recommendations that raise the clone comparison only to dismiss it — this is not 5 people complaining.

FindingThe complaint is that the game's build depth outpaces its resistance in the opening stretch — first wins arrive before players have explored the systems — and four of the five players say it as a caveat attached to a recommendation rather than a grievance.

• 80% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• Easiness recurs as a qualified aside, not a verdict: one player writes "Only complaint with the game so far: difficulty curve seems a bit on the easy side for my tastes", and another adds of the early chapters that "this is not necessarily bad."
• What they name is the opening stretch specifically — early chapters and the early curve — and no review describes the later game as trivial, so this is about onboarding pace rather than overall challenge.
• The sharpest version is the gap between apparent and actual resistance: the final boss "looked really daunting when i hit it in the tutorial but in my first real run i was able to beat it easily", from a player who calls the combo depth "astounding" and had "hardly scratched the surface."
• It is not unanimous even within five reviews — one calls the same stretch "a perfect (for me) difficulty curve through Chapter 5, not so easy its a pushover" — so part of this is taste spread rather than a flaw everyone hits.
• The only player who declines to recommend is the only returning one, and the only Chinese-language review: the predecessor was hard to work out solo and fully complete, while "this one, once you get the hang of it, is basically a steamroll." One instance is not a language-level rate, but comparison to the earlier game is what turns the caveat into a refusal.

Five mentions is a thin base, and four of them sit inside positive reviews, so treat this as a soft signal about early pacing rather than a ranked complaint.

FindingThere is no crash, bug, or lost-save complaint here yet — the theme currently holds one sentence from one review, and that sentence reports the opposite of instability. Its only grievance is battery drain on portable hardware.

• 100% of the reviews raising this still recommend the game — a majority-adjacent share, which makes it a tuning complaint rather than a dealbreaker

• The theme rests on a single English sentence from a single reviewer, so nothing recurs; there is no shared complaint to characterise, only one data point.
• That sentence names no defect the theme is about: “Runs well but does seem to suck up battery life.” It is a report of stability, not of crashes, bugs, or lost progress.
• The one real grievance is power draw — a performance and efficiency concern on battery-powered hardware, not a reliability one, and it belongs under performance rather than here.
• The reviewer still recommends the game and 100% of this theme's reviews recommend, so the mention reads as a caveat attached to praise rather than an attack.
• No save loss, no crash, and no progress-breaking bug has been reported under this theme to date; the honest conclusion is a null result, and the next Finding should be written from scratch once real reports arrive.
• All evidence is English and there is only one review, so no language-rate comparison is possible.

A count of 1 is below the floor at which any judgment about what players mean can be made — this is an empty theme, not a small one.