Sometimes game development goes wrong. It happens — technical problems emerge, designs can sound better on paper, early decisions can cause difficulties later on. It's just a reality of the craft that you will have failures and mistakes during development.
But what if you refuse to accept that a feature or project needs to be dropped or changed?
It's human nature to get attached to our work. In writing, we say you have to kill your darlings — those extraneous passages and phrases that you lovingly labor over that are ultimately superfluous or detrimental to the story.
And in game development, too, you must learn to cut the overambitious, the unworkable, the self-indulgent. If you don't...well, as we're about to see, it can drag a project down and maybe even stretch the company's finances to breaking point — especially if keeping something you recognize is bad means putting in significant additional work.
Economists call it the sunk cost fallacy, or the Concorde fallacy (in reference to the development of the Concorde supersonic passenger jet). It's an escalation of commitment where you feel that you've put so much money/time/energy into something that to quit now would be a waste of resources — when in reality your decision to continue should have nothing to do with that previous expense and everything to do with the likely worth of continued efforts.
We reached out to around a dozen developers to ask them for anecdotes about their experiences with the sunk cost fallacy in projects they worked on. Here are the stories they were willing to share. Hearing from them serves as an urgent reminder that all devs most be willing to perform a cost/benefit analysis at all times, and make brutal decisions based on the unpleasant conclusions.
Not seeing the forest for the trees
"GTA III had just hit and everybody was getting into like 'we want an open world game."
CloudGate Studios co-founder Steve Bowler remembers the costly effects of chasing a feature that never should have been there in the first place during NBA Ballers: Phenom development back when he was an animator at Midway.
"Open world had become like the new hotness," he recalls. "Like Grand Theft Auto III had just hit and everybody was getting into like 'we want an open world game.'"
The first Ballers game had been more straightforward — just one street-style one-on-one basketball game after another, each set in a different locale, with story cutscenes linking them together. Bowler says the sequel should have been a quick cash-grab project with a few new game modes and some extra stuff.
But they were instead determined to get a team that had never done open world development to somehow shoehorn it into an NBA-licensed streetball game.
NBA Ballers: Phenom
It was a disaster every step of the way. Once they got the driving system working, for instance, the NBA stepped in and said players couldn't run anybody over because the NBA is a family brand. Rather than drop it, since it was unnecessary to begin with, they put extra work in so that non-playable characters would dive out of the way.
"The team spun out trying to make open world work forever until they finally backed off and they made it isometric — like it's more like a walkable menu," says Bowler. "And even that I feel is ill-advised and they shouldn't have chased after it."
Glorious trainwrecks of lost productivity
Interplay founding member Rebecca Heineman says she could go on all day with sunk cost nightmare stories she's either witnessed or been involved in, but she points to just two of her favorites: "There's Interplay's Stonekeep, where they filmed hours of footage of actors in an outdoor stage," she recalls.
But nobody thought about the movement of the sun throughout the day, which meant shadows seemed to shift inexplicably from one scene to the next. "They believed they could fix the art in days," Heineman recalls, but it took 75 artists more than a year to go through the footage frame by frame.
"Renderware 4.0, which promised 'Next Gen' tools and features. The problem was that none of it was built yet."
A decade later she saw from afar as her friend Mark Dochtermann was blindsided by publisher meddling while trying to lead the technical side of Medal of Honor: Airborne development.
"The game was in development using the Ritual version of the Quake III engine, and [they] had to replace it with the Renderware engine because EA bought the company and wanted to use it company-wide," she explains.
Dochtermann clarifies that it "was in a console transition year and Renderware had sold EA on Renderware 4.0, which promised 'Next Gen' tools and features. The problem was that none of it was built yet so it was a glorious trainwreck of lost productivity for a long time."
After around 16 months with the Renderware engine, they finally gave up and switched to Unreal 3. They shipped a little over a year later.
Medal of Honor: Airborne
Never gonna give you up
The Banner Saga developer Stoic made the unfortunate mistake of twice falling into sunk cost nightmares, first with their console port and then with a now-cancelled Vita port. Studio co-owner and technical director John Watson admits to the error of thinking that, even after multiple missed deadlines by the porting company, they were "almost there" and so should persevere.
"That $100k sure would have been better spent somewhere else."
The full weight of this blunder became evident when the porting company went out of business about a year into the project. Watson says the failed console port cost around $100,000, and although some of that work was re-used when they restarted development with a different porting house, "that $100k sure would have been better spent somewhere else."
The Vita porting project was more complicated. It had initially been assigned to the first porting company in January 2015, then went on hiatus when that company folded in May.
"After the collapse and restart of our console porting effort, the new porting team was reticent to work on Vita," says Watson. "They felt that The Banner Saga was a big risk performance-wise and would require substantial GUI changes. They felt the development would be too expensive and lengthy to justify it. So we decided to pull the plug on that." But then Sony enlisted another porting house to take it on.
For well over a year this third team, Code Mystics, battled against complexities with the game's custom engine (which couldn't be tested until the port had reached a playable state). When they did get the game running, its poor performance on the Vita hardware led Stoic, Sony, Code Mystics, and Vita publisher Versus Evil to finally make the mutual decision to give up. They all wanted it to work, but Watson now concedes "that it was just not meant to be."
The Banner Saga
Ignoring the red flags
Veteran coder Glenda Adams remembers a troubled Mac game porting project in the early 2000s. It was for EA's Need for Speed: Porsche, and it came with "all kinds of red flags" right at the beginning. EA refused to relinquish some parts of the game's source code to her porting team at Aspyr.
"We eventually did realize it just wasn’t going to happen and pulled the plug, but it was agonizing to make that decision. And we put it off many months longer than we should have."
"We agreed to port the code we could, and have them put an in-house programmer on making a Mac library for the parts of the game we couldn't see," Adams explains.
For nearly a year, they struggled to make it work without being able to debug a large portion of the code (namely, everything withheld plus any systems that hidden code was intertwined with).
"We eventually did realize it just wasn’t going to happen and pulled the plug, but it was agonizing to make that decision," continues Adams. "And we put it off many months longer than we should have. In fact, we probably should have just skipped the project altogether when we saw the hurdles we were faced with."
"But there was optimism that we were at the top of our game, porting-wise, and we could make anything work."
Need For Speed: Porsche Unleashed
Puzzle Quest, Warlords Battlecry, and Gems of War designer Steve Fawkner has a similar story of failing to heed the red flags early in a developer-publisher relationship. Fawkner remembers a time when a publisher claimed his studio failed a milestone and withheld payment midway through a project.
"We started burning our own cash to chase the next payment," he recalls. This was seen as an investment to make future revenue, so it seemed rational. Then a month later, the milestone redelivered, and the publisher paid only half the money due and offered extra royalties as compensation.
"Our studio had been ruined, and we were down to three employees. We managed to survive and build back up again, and a very important lesson was learned."
"At that point, I should have realized what was happening and simply pulled the plug," says Fawkner. He knows now that it was a huge red flag that suggested the publisher was under financial stress. "[But] younger, naïve Steve said, 'Let's do this!' and plowed on!"
"Over the course of the next six months, the payments got smaller and further apart, but we kept chasing them," he continues. "The milestone failure reports got more nitpicky and bizarre, but we kept addressing them. We laid off staff, and we cut our wages, but we kept chasing those milestones because we’d now sunk more money into the project than the publisher!"
Worse, somewhere along the way, burned out on whacky change requests and budget strain, they had stopped playing their own game. "The game had gone from being a fresh take on something familiar and had become a weird Frankenstein thing that was not fun to play at all," says Fawkner. "We never noticed."
And when the milestone payments stopped coming altogether they still didn't stop working on the game for nearly six more months — when Fawkner finally made the call to cancel the project. "Our studio had been ruined, and we were down to three employees," he recalls. "We managed to survive and build back up again, and a very important lesson was learned."
Indies can also fall afoul of unnoticed or ignored red flags, as Voxel Agents creative director Simon Joslin learned with puzzle game Toy Mania. His studio put six months into developing it as a Facebook game — first with just him on the project, then also an artist and programmer. The warning signs that it wasn't working were there as early as the soft-launch, but Joslin decided to push on for another six months.
They collected data and tried to use it to refine the game and the experience around it (marketing materials, how it worked in different browsers, etc.) but then failed to heed the data's warnings that it still wasn't working. That led to further wasted effort, now with gameplay changes that made the main mechanic more difficult and less accessible.
"Then after months of failure on Facebook, we thought, 'Well clearly we need to be on mobile because this isn't working,'" says Joslin. They put a full team on a mobile version and spent a few months polishing it for an Android audience. "Then of course it failed there, too."
Some developers have found ways to avoid or lessen the impact of sunk cost fallacy. At his current studio CloudGate, for instance, Bowler says they usually swap tasks whenever somebody hits a problem that they're struggling with. That helps them to be "brutally honest" about what's not working and stay on track. At The Voxel Agents, Joslin has learned to follow his gut and design on paper before proceeding with cheap prototypes that have a clear vision to follow. That won't prevent sunk cost nightmares, but it will help.
Funomena co-founder and Journey producer Robin Hunicke has multiple stories of nightmare avoided due to cheap prototyping and an open and honest team environment. "Most of the things I have stories about ended up turning out OK," she says. "They just felt like a sunk cost for a long time."
At both thatgamecompany and Funomena, their blessing has been their small team size. When thatgamecompany pushed hard to put climbing mechanics in Journey, the team was constantly iterating on the idea in small prototypes, before moving the mechanics to full production too soon.
That would ultimately be to the benefit of the game, when the studio eventually gave up on the idea and switched to a scarf and flags system. "Same with climbing and stacking prototypes we did early on in Wattam, or the physical and musical feedback systems we designed for Luna," says Hunicke.
"I guess as a designer I really believe in trying things and learning from failures. Life is too short to hold on to bad ideas, but innovation requires grit and pushing on past the bad to the good."