Nearly a year into development of Defense Grid 2, Hidden Path Entertainment turns its attention to the most important and elusive element of making a game: creating fun.
I'm sitting in a restaurant Bellevue, Wash. with two of the founders of Hidden Path Entertainment and the lead designer of Defense Grid 2. It is late February 2014. We're here to talk about why Defense Grid 2 isn't fun.
Co-founder Michael Austin played the game for the first time last night. While he played, he sent his first impressions via email to co-founder Jeff Pobst. Pobst, who is also Defense Grid 2's executive producer, called for the meeting as soon as possible to share those notes with the lead designer John Daud. To say there was some measure of urgency would be an understatement.
Austin was the lead designer on the original Defense Grid. He's now designing Hidden Path's multiplayer exploration game Windborne. He took a rare break from designing that game to consult on the design for Defense Grid 2. He has detailed notes about his entire experience playing the game. He's making more, right now, in this restaurant, while we're waiting for the waitress to hand out glasses of water. He's pulling the details from his head. He's drawing on a piece of paper that looks like he stole it from somewhere. He's making a chart and bulleted lists — from his head — to gather his thoughts for the meeting that's starting in just a few seconds.
Daud looks nervous, even more so than usual. He watches carefully as Austin finishes methodically outlining his thoughts on the stolen paper. The list is of all of the things Austin thinks are wrong with Defense Grid 2. There are a lot of them. The chart is a visual aid.
The waitress steps away and the meeting begins.
He knows it's not fun. He's known it for weeks. What he hasn't known is exactly how to fix it.
The game, Austin announces, is not fun.
"The biggest thing, especially playing the early levels, is that I felt like I wasn't doing anything," he says.
Daud nods, silently. He knows it's not fun. He's known it for weeks. What he hasn't known is exactly how to fix it.
He's about to find out.
This is game development.
Just out of reach
The first time I set foot in Hidden Path Entertainment's office, Defense Grid 2 wasn't even a game, just lines of code and a collection of documents. That was April 2013. By late February 2014, the game has been in development for as near to a year as anyone can calculate.
Since my last visit to Hidden Path, in December of 2013, the Defense Grid 2 team has been in high gear, adding texture and detail to John Daud's level designs and overcoming various technological hurdles. Of the 20 planned levels, 18 are now playable versus the two that were complete as of my last visit. And most of the playable levels look finished, even if they technically aren't.
What began as an idea and a few thousand words in a document is now most certainly a game. The team has a few months left to make sure it's a good one. And they will need all of that time and more.
But even though the levels are coming together and the game is beginning to resemble what people will receive when they buy it, development has stalled on two technological fronts.
The first is the digital storefront. For Defense Grid 2, Hidden Path hopes to sell in-game items, but that functionality is waiting on code from Valve. And lately Valve has had other things to do. If the code doesn't come in, Pobst will have to move on and ship the game without the items. Maybe release them later in a patch. Maybe give up on the plan altogether. It will be a blow. Fans have expressed interest in the items, but the game must go on. Between "no items" and "no game," the answer is a simple one.
The game must go on. Between "no items" and "no game," the answer is a simple one.
The other front is more complicated. In spite of a year of development, the attention of two separate teams at Hidden Path and even a successful exhibition at the PAX Expo in September of 2013, Defense Grid 2's multiplayer still isn't working right. The PAX level was, by necessity, somewhat of a contrived experience. Players were playing head to head on a closed network using simplified code. After PAX, the challenge was to take that foundation and build on it and integrate it into Steam's matchmaking, which has not gone as smoothly as anyone had hoped. Players can find each other and start a game, but they can't play each other. Although the game will appear to be running, players can't even interact with it. It is, for all intents and purposes, broken.
For Daud, the multiplayer is his biggest headache. After over a year of planning and iterating on paper, he's finally seen the planned multiplayer modes come together — but he can't play them. He won't know what works and what doesn't until he can test, and he can't test until he can play. Multiplayer has been the one great uncertainty for Daud since before I first spoke to him about Defense Grid 2 in April of 2013. Now, almost a year later, the ability to put his own mind at ease on the subject remains tantalizingly out of reach.
"I know I can make a good game, a good single-player game, with what we've got right now," Daud tells me. "And I'm not sure ... Multiplayer is unknown ... [So] that is still a concern."
The process of getting multiplayer to work, of introducing it to the sequel for a game that didn't have it, has been a long one. In fact, the original Defense Grid was supposed to have multiplayer, too — the code is in the game; it just doesn't work. Multiplayer was a "nice to have" thing during that game's development but fell off due to time constraints.
This time it's not just "nice to have"; it's been advertised and exhibited. It's a planned feature. It has to work, or else it's a big disappointment. But Daud is going to have to wait just a little bit longer to see the multiplayer in action. Austin's notes are going to add to what's already a long list of tasks to complete, and as the team is forced to address them, multiplayer is going to slip even further.
The restaurant is called "Cactus" something. It's Tex-Mex. Daud chose it because the other option, The Cheesecake Factory, sometimes makes him feel ill. And for this meeting — best to take no chances.
Once the water glass and menu situation is sorted, Austin doesn't hesitate. There's no pandering to egos. It's just out with it: He was playing the game, and he was bored. He was so bored, in fact, he spent time zooming in on objects to stare at them and critique the artwork (he has notes on the artwork too).
"Let me start out by saying, I liked lots of things about it," Austin says. But he says he's going to be harsh, because he knows that most of the problems are actually his fault.
In Defense Grid, players build and upgrade towers using "resources," arbitrary units of in-game currency that accumulate at a variable rate, depending on how many resources you bank and how many enemies you kill. The better you are at the game, therefore, the better you can be at the game. And that was one of Defense Grid's hidden weaknesses. Players who struggled early on would still be struggling further along. The game was a blast for players who grasped early on how to maximize resources, but punishing for those who didn't.
"We kept leaving players behind that weren't getting it," Daud tells me privately the day after the lunch meeting. "It just compounded. If you don't do this one thing right, that's it."
For Defense Grid 2, Hidden Path wanted to make the game more accessible to newcomers, which meant making it less punishing to those who take longer to get good at playing it. That meant changing how resources accumulated. At the very start of development on Defense Grid 2, Austin proposed a solution: the drip.
Austin's proposal was to make resources accumulate at a fixed rate — a steady drip — of five resources per minute. Predictable as the sunrise. But the problem with predictability is it's ... well, predictable. Killing enemies no longer grants any resources and resources accumulate whether you're good at the game or bad. Equitable. Easy. Boring. The very tool the team put in place to make the game more accessible has also made it less exciting.
As Austin lays out his reasons, point by point, for why the game isn't good, there are winces — but also nods. They've been expecting this. What's surprising is Austin's stark admission that the drip, his drip, is holding them back.
"The other feeling that went with this is killing aliens didn't matter anymore," Austin says. "What I was doing was, I was staring at the resource thing and ignoring [the action]. ... Because it didn't matter whether my guns took out the wave or anything. All I cared about was, 'That's going up by five. I can make my decision in 15 seconds.' All of the drama was in the number, which is not where you want the drama.
"I actually ended up not paying attention to the aliens very much while I was playing. I was playing the numbers thing. Which is also not a desired outcome."
Austin's proposed solution is complex, but it amounts to a sort of compromise between original Defense Grid-style resource accumulation and the drip. It's basically a bigger drip that accumulates more resources faster along with an end-of-wave bonus that grants a flood of resources once a wave of enemies has been killed. In theory, it should combine the best effects of both styles of resource allocation, giving the player more opportunities to make decisions and be less dependent on the drip. In practice, this means more work — and time — from programmers who are already straining.
Daud has remained silent for almost half an hour, listening, taking in Austin's notes with stone-faced reserve. Finally Austin asks for Daud's opinion.
Daud's response: He's concerned that tightening up the resource system will take programming time away from making sure the multiplayer networking functions properly and that testing the new drip will take time away from testing the multiplayer.
"[T]here's a lot of stuff left to do," Daud says slowly. "We can make this a high priority and that's great, but something else falls off the list. That's my biggest concern. These are all great ideas. I want this. But I [also] want multiplayer to work."
"But this is critical," Pobst replies. "This is the fundamental rock that the whole game is built upon."
Daud says: "I know."
As the meeting wears on and chimichangas are eaten, Austin, Pobst and Daud outline the changes they'll need to make to make the game fun. There aren't many, but they are deep. And the most important is a tool Daud has needed for some time but hasn't asked for out of concern for the programmers' time, a tool that would allow him to load the game in between waves of enemies instead of having to play entire levels just to get to those waves. Now, in order to properly test and iterate on the new drip, that tool is essential.
As the list of changes accumulates, Austin and Pobst finally break and tell Daud to let it go and stop worrying about other people's time. His responsibility, they say, is to the game. Making demands on other people's time is his job. It's their job to figure out how to manage that. And Associate Producer Dacey Willoughby's job, specifically, is to make sure it can all get done.
"Why don't you let go ... and let Dacey do that?" says Pobst, exasperated. "Seriously."
Austin jumps in: "Your job is to ask her for unreasonable things and then let her figure out how to do them."
Daud nods. He swallows, and he says, "OK."
"Let me start out by saying, I liked lots of things about it."Michael Austin (center, back) arrives after the weekly team meeting to take lead designer John Daud to lunch.
The boss battle
Daud looks pale.
We're sitting in the conference room at Hidden Path. We've been back in the building for just over an hour. Now he's sharing Austin's news with the rest of the Defense Grid 2 team leads: Willoughby, programming lead Twig Johnson and art lead Lex Story. Daud is telling them how Austin and Pobst want the game changed — and about the work they will all have to do to make those changes happen.
Johnson has a list, handed to him a few minutes ago by Pobst, of the four things that need to happen immediately. All of them are unexpected, and all of them will take time away from other projects. The mood is tense.
I've been observing this team for over a year. I've been in rooms with them individually and as part of the larger group. We've shared meals, chatted. I feel I've gotten to know them as people, not just as interview subjects. But I've never seen them like they are now, discussing this list.
Everyone in this room knows that what Austin has recommended and what Pobst has officially asked for will change the game they have spent so many hours building and take away time they've planned to use for other things. Everyone in this room will be giving up something to make these changes happen. Decisions they've made are being moved aside or, in some cases, reversed.
Executive producer Jeff Pobst (far right) leads the level design meeting.
They are professionals, but they can't help but be defensive. And it shows. Here, in this room, these are not team leads or employees or even, really, game designers. In this room, right now, these four people are a team. And they are as unguarded as I have ever seen them.
The first item on the list: the drip. Double the rate — from 5 resources per second to 10 — and add the wave bonus. Not a controversial change, but it will halt work on almost everything else. Someone will have to go in and manually add data to the game code for every wave in every script for every level. It's tedious work, and it will take days, not hours.
The second item, though, is the one that will take the longest. It's the change to the code that will allow Daud to play a section of a level without playing the entire level. It will take someone days, perhaps weeks, but it will save Daud countless hours.
Johnson describes the change as "nice to have," almost dismissively and Daud cuts him off.
"Not 'nice to have' in a 'this will look awesome' [way]," Daud says. "Nice to have in, 'I don't have to play through the whole mission to get to the last wave and see if the change I made worked.'" The tension comes out in his voice, and it's not clear if he's taken Austin and Pobst's lesson about making demands to heart or if he's just frustrated. But the point is given and the team moves on. The change will be made.
Item three on the list is a change to one of the towers in the game. It's not behaving the way it should. An easy fix.
Item four is where the fighting starts. Item four is a change to the game's camera.
Currently the game camera does not allow players to zoom in tight to objects on screen. It's set to avoid collision with towers and buildings by limiting the degree to which you can zoom. Austin didn't like this. Austin wanted to zoom in to look at things. Mainly, he acknowledges, because he was bored, but he thinks not being able to zoom in is a problem regardless. He thinks players will want to be able to zoom in to see the action. His suggestion: Go back to the old way of zooming, the one from the original Defense Grid. The one Austin created in 2007. The one Story hates.
"No, no, no, no, no!" shouts Story. "We had that problem at the beginning. We opened it up and everything was out of frickin' scale!"
Daud suggests he take it up with Pobst.
"I'll let you guys fight," Daud says.
Story replies: "I'll win."
It's hard to say if he means it, but its also hard to mistake the frustration. The method of zooming the camera in Defense Grid prevented collision with tall objects, but it was a sort of visual trick, not a true zoom. Story thinks it's inelegant.
"It doesn't work for the game that we have now," he says. "We tried it!"
As the leads talk over why the change might be necessary at all, it becomes clear that camera collision is a problem they might not have given enough attention. Daud describes the many things he has seen clip through the camera while he's playing and it's an impressive list: tall buildings, the "hero" objects the art team has just created, core housings, some towers and, in many levels, the level terrain and towers themselves.
As they talk the problem through, the leads identify specific problem areas and solutions that might work for each. A plan begins to take shape. The idea is a compromise between not zooming at all and the FOV style of zoom Austin used for the first game, perhaps interspersed with additional technical tricks to prevent some of the bad behaviors that a camera clipping through objects might create.
The mood visibly lightens as the ideas begin to flow. What started as an affront to sensibilities is now just another problem to solve, another creative decision to be made by these people who make creative decisions for a living.
The four items on Johnson's list are added to the list they've already been working from, above everything else. Tasks related to those four items are assigned to various people not in this room. Conversations and clarifications are called for and scheduled. The process of game design, of middle management, of creation itself plays out and the machine keeps on humming.
As the tasks add up and the conversation turns to the next meeting, and the next, and then the next after that, Story sighs and says, "I don't like this Monday."
There's laughter, and then the Defense Grid 2 leads shrug off the tedium, set their minds on what's next and file out of the room to get back to work.
"No, no, no, no, no! We had that problem at the beginning. We opened it up and everything was out of frickin' scale!"
Willoughby, Daud, Story and Johnson (clockwise from left) gather to discuss Austin's notes and the new task list.
I'm sitting in a small meeting room at Hidden Path. It's a room just big enough for two or three people. I don't know what it's used for. The room is usually empty when I'm in town so they stick me in it. It's where I conduct interviews.
Today I'm interviewing Daud. It's Tuesday, the day after all the meetings. He looks tired but surprisingly, especially for Daud, upbeat. It's as if this phase the team has entered, the search for fun, is finally fully engaging him.
"[We] have people in the office who have been busy on other projects try the game out," Daud says. "So that happened basically for the first time last week. We had at least two of the designers take a crack at it. I'm actually getting feedback now, which is good."
Good or good-bad?
"It's good," Daud says, without hesitation. "At least you come up with some theories as to how to fix it. Those become action items. Then you wait for the tools so you can implement those things."
That's what Monday was about — an eight-hour whirlwind tour of what's been happening at Hidden Path for a week. Feedback, notes, action items. All geared toward finding the fun. Now the dust is settling and Daud is waiting for the changes to stick.
And then he'll do it all again. And again. And again. Until the game is done.
As usual, his biggest worry is multiplayer, which is still bogged down in technical issues. Meanwhile, the designs for each of the game's 20 levels are being polished and tweaked. Art, design and technology are finally coming together and the game is moving into the final phase: polish.
Art, design and technology are finally coming together and the game is moving into the final phase: polish.
For most of the team this means lots of meetings, fixing things that break or discovering problems as major systems are brought online, some for the first time. But for Daud, right now and again, it means waiting. In some ways he's back where he started, waiting for the game he designed to be complete enough for him to design it some more.
"I'll try and be patient," Daud tells me. "I have plenty of stuff to do in the meantime, but the longer that is delayed, the harder it's going to be. ... [But] we're still fine. We've done this before. ... I can't remember the last time that there was true panic in this office.
The story continues in Part seven.
This story is part of a series covering the development of Defense Grid 2. To read previous installments, please visit the Making of Defense Grid 2 page. This series will continue into mid-2014, the projected launch date for Defense Grid 2.