clock menu more-arrow no yes
Graphic artwork featuring a montage of characters from the video game Phantasy Star Online over a dark grey pixelated background
Phantasy Star Online artwork
Graphic: James Bareham/Polygon | Art: Sega

Filed under:

Phantasy Star Online programmers on breaking new ground and their Diablo-style isometric prototype

We continue our retrospective interview series for Phantasy Star Online’s 20th anniversary

With 2020 being the 20th anniversary of Sega’s Dreamcast RPG Phantasy Star Online, we’re looking back with a series of interviews featuring key team members who worked on the game. Over the past four weeks, we’ve posted chats with producer Yuji Naka, director Takao Miyoshi, composer Hideaki Kobayashi, and script writer Akinori Nishiyama/effects designer Takanori Fukazawa.

Now we’re digging into the nitty-gritty with a roundtable featuring three original team members — lead programmer Akio Setsumasa, character programmer Yasuhiro Takahashi, and enemy programmer Masakazu Miura.

Younger gamers who have grown up with the always-on, connected nature of today’s consoles won’t necessarily remember the excitement, aggravation, or challenges of connecting a Dreamcast, PlayStation, or GameCube (or Saturn, if you were really ambitious) to the internet on a dial-up modem. With Phantasy Star Online, the designers and programmers at Sonic Team found themselves in uncharted territory.

For instance, there was no way to patch a console game in those days. Sonic Team had to try to get everything right before it committed the game to disc, and was only able to fix mistakes with the release of Phantasy Star Online Episode 2.

Read on for revelations of the game’s darkest secrets, including the infamous “black screen of death.”

Hello, since there are three of you, it would be great if you could introduce yourselves and describe your individual roles on Phantasy Star Online.

Akio Setsumasa: My name is Setsumasa and for Phantasy Star Online, I was in charge of system engineering-related tasks such as servers, communication, server management, and quest systems. On top of that, I also did quite a bit of game design, in terms of planning the gameplay.

Yasuhiro Takahashi: You were involved from the prototype stage, correct? Before a game is made, there’s an experimental stage — figuring out which game mechanics are going to work, etc. — and Setsumasa was involved from that stage.

Setsumasa: Mostly, I did server management and operations. For the Blue Burst version that was released for PC, I was the director on the project.

Takahashi: I am Takahashi and I joined the project after most of the base systems were established, and I ended up working on a lot of the character aspects of the game. So, designing the protagonists, the bosses, and the enemies. I worked on the game select screen and world select screen, too.

Three former Sega programmers stand in front of a black, tiled wall
From left to right: Akio Setsumasa, Masakazu Miura, and Yasuhiro Takahashi
Photo: James Mielke for Polygon

When you say you were working on character designs, do you mean how they would function or the actual visual design?

Setsumasa: To be more specific, I worked on the enemy AI, how the characters are animated, and so forth. And how you can design your character, adjust their height or color.

Masakazu Miura: My name is Miura and I joined the project around the same time as Takahashi. I was responsible for enemy designs; both regular enemies and boss enemies. And when I say “designs,” I mean their AI and so forth.

We’re now approaching the series’ 20th anniversary. You were all 20 years younger when you started working on the game. Looking back, what are your feelings about PSO, now that you have the benefit of hindsight?

Setsumasa: We’re all still currently supporting Phantasy Star Online 2, and we worked on the various PSO games that came between PSO and PSO2, so it feels like 20 years crept up on us. Working on it every year, it feels like it’s been short and long at the same time.

Takahashi: It’s hard to tell if we’ve changed at all.

So it’s not like you worked on PSO and then there was a void of 20 years. Phantasy Star Universe, Zero, Nova, Portable, etc., all filled the gap.

Setsumasa: Yeah, it feels like it crept up on us.

Miura: It’s our life’s work.

Takahashi: I was involved in designing the characters, so during development we were very concerned and focused on how to make the best experience for players, because it’s a network game so we had to design the game in a certain way. We were wondering how we should design the game systems, and how we should put all the elements together to make a fun experience. So even for me, individually, it was a tough process in terms of development, but I learned a lot through experimentation.

Miura: For me personally, I had no idea 20 years ago what a network game was supposed to mean, and I was primarily concerned about making a game that players would enjoy. But I had no idea that PSO would become this big and that we would expand the PSO IP to the extent that it has today. So the fact that [you came all the way to Sega in Tokyo] to ask about PSO is something that makes me very happy.

Can you describe what the first meetings were like? Was it — according to legend — originally called Third World before changing to Phantasy Star Online?

Setsumasa: I don’t remember the original title, but it wasn’t Phantasy Star-themed. It was originally supposed to be a new IP with an original universe. It was scheduled to be a sci-fi RPG, but the specifics of the game world had not been decided.

For the team’s first online game, why did you all decide to make a science fiction game versus a more traditional fantasy game?

Setsumasa: Hmm. It just happened to be a sci-fi game. I’m not sure. [...] At the beginning there was a city — the city you see in the final game — that was designed first. That area showed the feeling we were going for and that we wanted to reproduce on the Dreamcast hardware. I believe it grew out from that original concept.

Is that the hub city in PSO?

Setsumasa: Yes, pretty much the same city, as it was originally conceived.

So the city naturally evolved and organically became PSO, as we know it?

Setsumasa: Initially, the programmers were working on testing the game mechanics, and the artists were off making character designs and the environments. But they were struggling to come up with those designs, so [producer Yuji] Naka suggested using the Phantasy Star universe, and the team was OK with that. Once it was decided to use the Phantasy Star universe, things really started to come together in terms of the visual aspect of the game.

Early preview builds of the game showed the Diablo influence. You’d clear out areas and hope that the enemies would drop cool items. Was the team assigned to play Diablo inside and out?

Setsumasa: Yes, I played Diablo quite a bit. There was only Diablo and Ultima at the time, for network games that we could use as references. The whole concept started with making a game like Diablo, but when showing it to the artists, they didn’t think it was artistically appealing and thought it was too dark. So I knew that we couldn’t just make a Diablo game. It had to be catered to our market. So we put our heads together to figure out how best to bring a Diablo-like game to Japanese consumers.

Takahashi: We played quite a bit of Diablo in the office, didn’t we? I remember that we had Diablo and Ultima Online installed on the PCs in the office, and we had play sessions where we would watch people play Diablo. It was through watching Diablo and Ultima in the office that we saw the appeal of playing online and networked games.

Setsumasa: I tried to get people to play [these games] to understand the appeal, but their responses were pretty lukewarm. [laughs] So I was forced to think about how to make PSO interesting to the rest of the team, and ultimately, to the consumer. [...]

To go back even further, at the point where it was decided that we were going to make a game for the Dreamcast, we decided that network games were going to be the next big thing, so we requested the inclusion of a modem in the hardware. We knew that if the modem was an optional peripheral, nobody would play the game.

If the modem was optional, you would have had a split user base, so it was integral for everyone to have the same hardware. Diablo is played from an isometric viewpoint, but PSO is played in the third-person, which I find a lot more attractive because you can see the details of your character better. Was this a conscious decision?

Setsumasa: We did make an isometric prototype of PSO, but there wasn’t enough action and excitement in the gameplay. And there are big dragons in the game, and our team talked about wanting to be able to look up at the dragons from the perspective of the protagonist, so we decided it would be better to make the game from a third-person view.

ChuChu Rocket! was Sonic Team’s “tech demo” to test the Dreamcast’s network capabilities in terms of funneling data, but PSO demanded a lot more out of the online play. What were the brainstorming sessions like? Were there lots of ideas that didn’t make it into the game because of the hardware limitations?

Takahashi: We were all part of Sonic Team so we were able to gain the necessary network knowhow through ChuChu Rocket!’s development. Through ChuChu Rocket! we figured out how to offer four-player online multiplayer gameplay, so we were able to use that experimental knowledge to build PSO. It’s because of ChuChu Rocket! that we realized we could make a game that offers online communication between four players.

At the same time, the team also figured out how to build 3D games through their experience with Sonic Adventure. Taking these two experiences, we figured out our capabilities and limitations, and we were able to communicate this to the planning team early on. That really helped us to lay out the scope of the project from the beginning, and we had a smooth start. Although maybe it’s just that my memory fails me. [laughs] It may have taken us some time to convince the planning team of the limitations, but Setsumasa was in those early concept meetings, so he was able to tell the planners what we could and couldn’t do. There weren’t really any problems. Especially since we had a prototype to refer to.

Setsumasa: There were also scope discussions about making the game so that it could be played worldwide.

Takahashi: A lot of the conceptual ideas came from Naka, [director Takao] Miyoshi, and [art director Satoshi] Sakai. But when it came to the technological aspects, it was our part of the team that was doing the grunt work. In terms of our aspirations at the time, we wanted to make an online console game that we could sell and connect people to on a worldwide basis. [to Setsumasa] Do you remember what those early conversations were like?

Setsumasa: The biggest difference between PSO and PC games was the fact that we were making an online game for the Dreamcast, so a lot of those early meetings that we held were focused a lot on how players were going to be able to communicate with each other effectively, because that’s a big part of online games.

Unlike PCs, the Dreamcast didn’t come with a keyboard, so we had to figure out ways for players to send each other messages. We had a few different systems. We had the Word Select Screen, where you could select between different words and phrases. We had the Symbol Chat Screen as well. We tested all of these features out, and a lot of them did make it into the game. The focal point of a lot of these early meetings was how players were going to effectively talk to each other with alternative methods to the keyboard.

Another feature we discussed from early on was how to simplify the process of getting players online. Because that was still quite complicated back then.

Miura: There was a lot of discussion about how to best bring an online game to the home console.

The Dreamcast keyboard was already out by the time PSO launched in late 2000/early 2001.

Takahashi: I believe the keyboard was already out when we released ChuChu Rocket!.

Setsumasa: Yes, it was.

Takahashi: But we couldn’t assume the consumer would own a keyboard, because the Dreamcast came with only the controllers. So we had to make sure that, at the minimum, you could chat using a Dreamcast controller.

PSO has a really elegant chat system utilizing icons to communicate. It’s impressive that you were able to compress so much info into such a limited space. How big was the development team overall?

Setsumasa: I think we had 10 programmers on the team. Give or take.

Takahashi: Three on servers. How many in planning? Another 10? And 15 artists? A couple people for sound. All together about 40? Does that sound right? Maybe better to check the final credits on YouTube. [laughs] But, 50 people max.

Setsumasa: No, I don’t think there were 50. It’s important to remember that PSO wasn’t the only project we were working on at the time. So there was a lot of overlap with other projects we were working on.

Nowadays, 50 people is a modest team. You look at Call of Duty, and they have hundreds of people, if not thousands. To accomplish what you did — in basically uncharted development territory — with only 50 people is very impressive.

Setsumasa: It was a big team back in the day. Actually, Sonic Adventure had a lot more staff on their team. It does seem modest compared to the team on Sonic Adventure.

A winged monster screams
Phantasy Star Online artwork
Art: Sega

Especially because Sonic Adventure seems like a much more traditional game experience. Besides the cutscenes, you’re basically running forward most of the time, and there weren’t all these networked multiplayer hurdles to overcome. It’s more of a traditional game experience. PSO was entirely unique.

Setsumasa: That’s true.

What were the most challenging technical aspects of the game?

Setsumasa: As mentioned earlier, we were all on the Sonic Team, so we had a lot of experience with action games, so I was pretty confident that we could pull off the action portion of the game. It was really the network portion that concerned us. I wasn’t sure how we were going to keep players online while playing an action game without major latency.

Takahashi: We spent quite a bit of time on the packets.

Setsumasa: Yeah. The digital communication of the development was the most challenging [aspect], because this was before high-speed modems. We had to use the Dreamcast modem in order to communicate, so we had to figure out how to have the least amount of lag time while playing the game.

Takahashi: We had to figure out how to reduce the number of packets that were being transmitted by the hardware in order to reduce data, and calculate how much data players were using, etc. Because if there’s too much lag in the game, an action game wouldn’t be fun, so we had to figure out how to minimize the amount of packets the game was broadcasting. The packets also determined the characters’ movements, because they had to move in ways that would minimize data transfer.

Setsumasa: So we had to figure out how to transmit the packets, [and] how to communicate as infrequently as possible while allowing for an experience that did justice to action games, so that players didn’t feel a sense of incongruence when playing the game.

The characters ran at a standardized speed, and you couldn’t run and shoot at the same time; you had to stop and shoot. You can’t jump in the game, either. Was this a very deliberate design choice to reduce the amount of data that had to be sent back and forth?

Setsumasa: Yes, those design elements were due to the limited capacity of the packet transfers, and that was something that I made clear to the planning department at the beginning of the development. We expressed which actions were and were not possible with the intent of putting the game online.

Takahashi: These were things that we were able to implement at a later date, but initially, anywhere where there was an on/off switch, like the command for opening a door, the command could only be switched on once and couldn’t be turned off. So, in the example of the door, you could open the door but not close it. The reason being that there was no guarantee that the command would be delivered through the network in time to sync with other commands. So, you could send the command to open the door, but we couldn’t keep track of another person possibly closing the door, so once the door was open it had to stay open.

So, back to the point of the characters’ movements and design, those were limitations due to the game being on a network. And the on/off switch limitation was another example of where the game was restricted due to being an online game.

Setsumasa: So the parameters we presented to the planning team were quite granular from the start.

It’s really amazing from a design standpoint, how you had none of the modern technical advancements we have today — high-speed broadband, enormous memory storage — and still you made it work. The gameplay never suffered from not being able to close a door. There wasn’t really a reason to have to close the door; monsters didn’t cross those barriers. It was a very elegant experience that people don’t realize was designed out of necessity.

On a side note, when you’re creating your character, you can make a big character bigger, or take the smallest character class and make the tiniest version of that. Did that expand or contract the hitbox of the character, and were there any interesting anomalies that occurred as a result? For example, if you were a tiny character, could you exploit the weakness of a large monster? Anything like that?

Miura: The visual attributes of the characters were purely cosmetic, and their abilities were identical. The reason we did this was if there were advantages to being visually different, we assumed that players would all choose the character that was most advantageous. Even the hitboxes were the same, regardless of the size of the character. We wanted players to design their characters purely on how they looked, without considering the consequences of those choices. However, there were minor issues — you could call them bugs, I guess, or unintended consequences. Depending on the height of the character and the guns they held, the aim and target would differ slightly. Other than that, the characters functioned the same way.

PSO was intended as a multiplayer game, of course, but it also supports a single-player mode by allowing offline play with the same character you use online. It was nice to be able to enjoy the game without going online, but the single-player experience was more or less identical to the online experience, minus allies, which could be quite tedious. Was this your original intention? Were you aware of the limitations of playing single-player?

Setsumasa: Yes, the balance in the game was the same for multiplayer and single-player.

Takahashi: Did we adjust the parameters? I feel like I remember making a table [to adjust for single-player].

The enemies all had very predictable routines. You could run in and attack or shoot at them, and they would all come sliding toward you. So the standard practice was to get a couple of hits in, run out of the zone, and wait until they walked away again, then repeat. It was a little cheap, but there was no other way to really deal with crowds. This was really the only way to survive in single-player.

Setsumasa: You’re right. PSO was very deliberately designed for both single-player and multiplayer. When playing in multiplayer, you could clear a single stage or battle faster as a group than playing alone. But, we did make it so that if you went back into an area, you could shoot the enemy from afar. Otherwise, given the way the game was designed, you would get killed. One of the objectives for us was that the gameplay wouldn’t change too drastically between modes. We tried to make a unified experience. But, we wanted to give the players some variety in how to approach the game, so we also had difficulty levels — normal, hard and very hard — so that players could choose the level they were comfortable with.

In creating the loot system, how did you decide how many items you would put in the game?

Setsumasa: We initially decided on the total number of items and their ranks, strengths. Then we calculated how many unique images we would need, and that was limited by the artist’s capacity to create the graphics. So whatever the artist was able to produce within the development cycle capped the number of items we could put in the game.

Takahashi: Come to think of it, I don’t think we maxed out the GD-ROM’s storage capacity. Do you remember?

Setsumasa: In terms of memory space, the graphic data didn’t take up too much space. It was the sound data that took up a lot of space.

Miura: It’s always the sound data. [laughs]

And you couldn’t patch anything afterward, so ...

Setsumasa: No, nothing like that. It was difficult deciding on the appearance rate of the items. We wondered if we set an item to appear once every millionth play, would it ever appear if there weren’t enough people playing the game? We had to use our best judgment [to balance the rates at which items dropped]. After the game was released, it was a relief when we heard from people playing the game that a certain rare weapon had dropped, because there was nothing we could do if it didn’t work. We just had to have faith.

Takahashi: There were rumors about certain weapons not dropping. As a programmer, I would have liked to have patched the game, but there was no system in place to send patches. There were talks of using memory cards as a way of distributing patches, but the issue was memory caps on memory cards, and the number of times it would work and where it would work, etc.

Miura: Didn’t we have a download quest?

Setsumasa: Ah, yes. We did do a download quest. In PSO, you could download the script and data for a new quest onto your system, and if you completed that quest, you would get an item.

Miura: If we got a complaint from a user that a certain item wouldn’t drop, we set it up so that they could download this “Download Quest.” By completing a quest, they could get a particular item. We couldn’t patch the data or fix an issue that a player encountered, but as a means of customer support, we could deliver a particular item to them.

In regards to the rarity of the game’s best weapons, because the red boxes were the rare drops, what was your algorithm like to determine what the drop frequency was? Was it, like, 0.05% or something? Is there a specific item that is the rarest drop?

Setsumasa: We laid out in an Excel sheet all the items and roughly how frequently they would appear per X number of plays. We didn’t do a simulation of the number of times it would take for an item to drop; we just punched out the numbers depending on how frequently we thought it would be appropriate.

Do you remember the rate of the absolute rarest item?

Setsumasa: I can’t remember the name of it, but there was one that was super rare and only a handful of people would ever get it.

Takahashi: I think there was something in Blue Burst.

In English, we called a particular phenomenon with PSO the “black screen of death,” aka “BSOD,” which happened if you lost your network connection while you were loading into another area. Nine times out of 10, your game data would be corrupted and you’d lose everything. Since player data was console-stored, not server-stored, there was no such thing as data recovery. How did you guys handle that?

Setsumasa: This was a really serious issue for us. I can only apologize. For the Dreamcast version, we released a version 2.0 shortly after the initial release. Since we couldn’t patch the game, we had to release an entirely new version. And we shipped version 2.0 immediately [in part to deal with this issue]. In addition to fixing bugs, we added some content as well. I remember we did that in a really short period of time, and it was a difficult time.

Takahashi: We included the bug fixes, because that was the point, but we included new content to convince the users to buy the new version. I wonder how long it was before we had version 2.0 available?

Setsumasa: It was within six months. Also, there were a few issues with the server that we were able to circumvent from our side using backdoor fixes. I recall that those were quite difficult to fix.

A Dreamcast sits inside a glass case, looking like a museum exhibit
A Dreamcast console on display at Sega of Japan
Photo: James Mielke for Polygon

Were there server issues in version 2.0?

Setsumasa: No, there were a few in the first version. And for version 2.0, we fixed the server-side things. There were similar issues with the GameCube version as well. That’s why we released Phantasy Star Online Ver. 2 for Dreamcast and Phantasy Star Online Episode 1 & 2 Plus for GameCube. Again, patches didn’t exist, so this was the only way to fix bugs in the game.

Miura: It was quite serious. But those were the limitations with the technology at the time. The system was developed to use the same data for online and offline play, using the same memory card. And the memory card save data was problematic in and of itself. It was prone to critical errors under normal circumstances. I still feel bad for the inconvenience to users.

Setsumasa: The crash didn’t delete the save data, just the equipment data. This was actually part of an anti-cheating mechanism that was in place so that players couldn’t tamper with the equipment data. There was another game released around the same time — not from our company, but a game that had a similar bug — and the network crash would lead to duplicating the equipped data. So, this led to players “duping” the equipped items and tipping the balance of the game, basically killing the game because weapons were all collected. We decided that it would be better if the equipped data disappeared rather than increasing. That’s why the items disappeared when the game crashed, which was regretful for the player.

Takahashi: It was supposed to be a tactic to prevent people from exploiting the game during loading. In other words, it was intentional that the equipped items disappeared.

Setsumasa: Yes, it was intentional.

Miura: There was a loophole during loading that made it possible for players to cheat the game. We temporarily removed the equip data during loading and returned the data after loading. With this bug, the game reverted back to the save data without the equip data. But without this system in place, the equip data would multiply every time the game loaded.

Setsumasa: So, the root of the bug was placed there to prevent cheating the game.

Miura: Yeah, that’s why the bug was inevitable.

Setsumasa: We put in a lot of preventive measures to keep people from cheating the system, so when this bug was discovered, I regretted that we had possibly gone too far.

With the GameCube and Xbox versions of PSO, did their advanced hardware in relation to the Dreamcast enable you to do anything under the hood to improve the game?

Setsumasa: The GameCube system had online compatibility, but it wasn’t a default accessory with the console, so we weren’t sure how many people would play online. We assumed the worst scenario — that not a lot of people would play online — so we added the four-player split-screen multiplayer in order to give GameCube players another option to play. GameCube was also superior hardware, which made the four-player split-screen possible. Originally Nintendo was going to release the GameCube with a 56K modem only, not the broadband adapter. When Sega announced that PSO was going to be released on GameCube, Nintendo received a lot of requests from consumers to offer a broadband adapter.

The Xbox version, even though it was running on the most powerful console at the time — which also had an internal hard drive — ended up being the most hacked version of PSO.

Setsumasa: There was a request from Microsoft to utilize as many Xbox-exclusive features as possible. I think we also included voice chat.

Why did you choose GameCube over PlayStation? Nintendo was pushing back on online games at the time, and considering its views on network games ...

Setsumasa: When we were comparing PlayStation 2 to GameCube’s capabilities, it turned out that PlayStation 2 wasn’t as powerful, and there were a lot of things that made it difficult to develop on PlayStation 2. For example, we could only use 16 colors for textures. When doing a side-by-side comparison, it was easier to port to the GameCube over the PlayStation 2.

Which of the bosses was the most difficult to design?

Miura: The De Rol Le dragon.

The dragon that swims around the boat?

Miura: Yes.

Of all the Japanese MMO-style games after PSO, which one do you think inherited the spirit of PSO the most?

Takahashi: Monster Hunter.

Setsumasa: Yes, Monster Hunter. The main programmer on Monster Hunter even said himself that he played a lot of PSO. We’re actually quite close, and actually have used Monster Hunter as reference for making the newer PSO games, so the influence goes both ways.

Nowadays, a device like the iPad seems easily as powerful as the GameCube was back then. Has the idea ever come up to port PSO to modern devices? It seems like it could coexist peacefully with PSO2 because it’s more of a nostalgic experience.

Setsumasa: There’s the question about how to port the controls to a tablet.

The Apple TV has a controller.

Setsumasa: True.

There’s an incredible desire to play the original without having to have a Dreamcast or GameCube. People are really hungry to reexperience that world, but the obstacles to do so are a hassle to work around.

Setsumasa: Bringing back the Blue Burst servers is probably the easiest way. It’s true. It would be nice to do so.

What do you think would be the best hardware? Switch?

Setsumasa: Sure. I would like to put it on the Switch. There’s a feature on the Switch called “cloud gaming,” where people stream the game instead of storing the game on the hardware. I think that would be an option.