A studio listed 10 reasons not to buy its own game. It shot into Steam's top 10

The games business has one problem underneath most of its others. You cannot tell whether a game is finished, whether it runs, or whether it will still be switched on next year until after your money has gone. Everything the industry has built on top of that - refunds, demos, review scores, early-access labels, streamers - is a machine for making quality visible before payment. This week every one of those machines was visibly straining, and one studio tried the strangest fix available: telling people not to buy.

The advert that argued against itself

Developer Bulkhead published a 15-minute video titled “10 Reasons NOT to Buy” its own shooter, Wardogs, ahead of a Steam early-access release on 10 September [1][2]. It is not a joke premise. The list is real, and it is unflattering.

Point by point: there is not much content yet, and there will not be regular content drops [1]. The 40-dollar early-access price is steep, and it will rise to 60 dollars by the end of early access [2]. Player numbers will fall over time [1]. There will eventually be some form of paid extras, though not during early access [2]. Performance and visuals are not there. Controller support is thin. And reason seven is simply the name of the studio’s last shooter, Battalion: 1944, which failed [2].

Bulkhead said it wanted to do “something most AAA marketing teams would tell their dev teams to never do” [2]. On price it was blunter still: “If the price sounds too high for an early-access game, please don’t buy it yet” [2].

Within 24 hours Wardogs was in Steam’s top 10 best sellers [1].

That is not a paradox, and it is the whole point of this briefing. Anyone can claim their game is good. Almost nobody can afford to publish a list of its faults. So the list itself is the evidence - it is expensive to send falsely, which is exactly what makes buyers believe it.

Four million people bought the deluxe edition of a game nobody has played

At the other end of the same problem sits Grand Theft Auto 6. The analytics firm Sensor Tower found 89% of pre-orders are for the 100-dollar ultimate edition, not the 80-dollar standard [3]. The split is near-identical on both platforms - 90% on Xbox, 88.5% on PlayStation [3].

That is far outside anything the industry normally sees. Sensor Tower’s Karl Kontus said typical premium-edition shares run around 10-20%, and a 50% share during pre-orders would already be a great number. “I have never seen a 90% premium share before,” he said [3]. The firm’s tracker had the game at 4.3 million pre-orders when the figures were published [3].

Nobody outside Rockstar has played it. The most money is moving at the point of least information, which is the reverse of how you would design a market if you were designing one.

Take-Two boss Strauss Zelnick has called the pre-order level “unprecedented and astonishing” [3]. He also said nobody yet knows how it will translate into sales [3]. Kontus reckons the eventual figure lands nearer 25 million pre-orders, roughly 2 billion dollars banked before launch day [3].

Meanwhile the group claiming responsibility for this week’s fresh GTA 6 footage leak published a manifesto attacking exactly this. Among its demands: end digital pre-orders. “If publishers want revenue before launch, they can press discs,” it wrote [4]. Take the claims for what they are - an anonymous group with an obvious interest in attention. The complaint is still the same one: money first, information later.

The signals we built for this are all a bit crooked

Steam reviews are the industry’s main quality signal, and a new study suggests they measure something odd. Researchers Chetan Mudlapur and Om Prakash Singh at the Indian Institute of Information Technology analysed over half a million Steam reviews [5].

The pattern splits three ways. Reviews from players with 0-2 hours skew negative - that is the refund window, so those are mostly people who bailed and got their money back [5]. Between 2 and 200 hours, reviews are consistently positive [5]. Past 200 hours, negative reviews spike sharply again [5]. Around 50 hours is where players are most likely to leave a good one [5].

The researchers point at sunk cost: people who have poured hundreds of hours in struggle to walk away even once the game has changed into something they dislike [5]. Genres with open-ended, service-dependent designs showed the highest rate of this split; story-driven games with a clear ending showed the lowest [5]. So the loudest voices on a store page are disproportionately the ones who quit in the first two hours and the ones who cannot bring themselves to quit at all.

The discovery layer is thin too. Bain & Company surveyed over 5,300 players and found two-thirds prefer familiar games or sequels, with only one in five actively seeking out new titles [6].

And where a real signal exists, players pay for it. About 600,000 people spent 20 dollars extra on a Premium Pass to play Halo: Campaign Evolved five days early [7]. Roughly 2.1 million played it through Game Pass on day one [7]. Alinea Analytics estimates the game sold 1.2 million copies in two weeks and grossed 67 million dollars [7]. Paying 20 dollars to find out sooner is a price on information, not on the game.

Even the words on the box are a signal that can be tuned. Former PlayStation boss Shuhei Yoshida says he told the makers of Clair Obscur: Expedition 33 to avoid the phrase turn-based role-playing game [8]. The label alone, he thought, would make players skip it [8]. Sandfall Interactive instead marketed “reactive turn-based battles” [8]. The game has sold more than eight million copies [8].

Where the signal is what pays, it gets gamed. Some Marvel Rivals players have been dropping config files into the game folder mid-match to trigger a cancellation and dodge a recorded loss [12].

What it costs the studio that is genuinely good

If buyers cannot tell, the good game gets priced like the average one - and the maker eats the difference.

Arco, a Mesoamerican fantasy role-playing game, passed 100,000 sales this week and finally recouped what it cost to make. It launched on 15 August 2024, so that took two years [9]. Composer and sound designer Jose Ramon called the milestone “incredible” [9].

Planet of Lana’s team pitched it “hundreds” of times and were rejected “hundreds” of times before one publisher said yes, after three years of unpaid work [10]. “It sucks to be told no,” co-founder Adam Stjarnljus said of the run [10]. Bain’s data points the same way. Of 100 titles released since 2023, 83% of games built for a specific, describable player found commercial success, against 50% of broad ones [6]. A game that is clearly one thing is easier to believe.

Trust runs the other way too. Sony’s plan to end PlayStation disc production from 2028 has produced a slow revolt [11]. Its Phantom Blade Zero livestream this week had its chat flooded with “no disc no buy” [11]. The developer, S-Game, had nothing to do with the decision [11]. Britain’s Digital Entertainment and Retail Association called the move “a triumph of corporate convenience over consumer choice” [11]. Konami will switch off Castlevania: Grimoire of Souls on 3 September, five years after launch, leaving it unplayable [14].

Some publishers are treating that as their signal. Tripwire Interactive is building community server tools for Deceive Inc so the game keeps working after its developer, Sweet Bandits, shut down in 2024 [13]. “Live-service is a red flag for us,” publishing director Bryan West told PC Gamer [13]. Pearl Abyss reported a quieter quarter after Crimson Desert’s launch surge [15]. Its accounts still carry “revenue deferrals for refund guarantees” - money held back in case buyers ask for it back [15]. And the commerce firm Xsolla launched tools aimed squarely at what happens after the first purchase [16]. Studios get their own shop and their own player data, rather than renting both [16].

The money moved anyway

Alibaba is selling its games arm, Lingxi Games, to private equity firm Trustar Capital. Reuters reports the company should reap more than 2 billion dollars, citing a person familiar with the matter [17]. Bloomberg, which first reported the internal staff memo, put the deal at least 1.5 billion dollars [18]. Both figures are from sources, not a filing. Lingxi’s chief executive Zhou Bingshu and his management team stay in place [17].

Take-Two’s 2K opened a new Vancouver studio, Small Axe Studios, to build a sports franchise [19]. It is led by Aaron McHardy, after nearly 18 years at EA on Madden, College Football and FIFA [19]. Mattel, which owns Barbie, Hot Wheels and Uno, folded roughly 300 game workers in Los Angeles and Hangzhou into a new Mattel Game Studios [20].

And the people

Xbox developers held Save Our Devs rallies in nine cities across the United States and Canada on Tuesday [21]. The Bethesda Game Studios union and the Communications Workers of America backed them [21]. Around 1,600 people have already gone from the Xbox group, with a similar number of cuts expected again [21][23]. A Bethesda developer told Kotaku that cuts were rare before Microsoft’s ownership: “All of a sudden we have yearly layoffs” [21].

The week before, Xbox chief executive Asha Sharma visited Bethesda’s Rockville office [22]. Union members raised one red flag for each laid-off developer and parked a giant inflatable rat on the lawn [22].

Inside the same company, Blizzard closed its 2025-26 financial year as Xbox’s top-performing studio, with its third-best revenue ever [23]. That comes from an internal email by president Johanna Faries that leaked to Windows Central, not a published figure [23].

Netflix closed Night School Studio, maker of Oxenfree and the first studio it ever bought, along with Helsinki’s Moonloot Games and an undisclosed number of in-house roles [24]. Night School’s horror game Unhinged had shipped about six weeks earlier, and co-chief executive Gregory Peters had praised it on the company’s second-quarter earnings call [25].

One quieter note to end on. Free Lives will take its office beat-em-up Stick It to the Stickman out of early access on 23 September [26]. It spent a year and a bit with players buying it unfinished and saying so [26]. That is the honest version of the deal Wardogs is now proposing. You pay early, you are told exactly what you are paying for, and the finished thing arrives when it is ready. As one industry writer put it this week, the market increasingly rewards launches that feel like events rather than a steady drip of safe sequels [27].