Shroud of the Avatar: Richard Garriott Talks Selective Multiplayer

Richard Garriott met with Warren Schultz, from About.com’s Game Industry portal, for a follow-up interview about Shroud of the Avatar. The topic, this time around, was Shroud’s “selective multiplayer” features…and in it, we learn that Richard Garriott still very much hopes to bring the game to mobile devices one day:

Garriott On selective multiplayer: We were trying to concurrently solve a handful of problems. The problem we were trying to solve first was my demand to be able to play on a mobile device when I’m not connected to the internet. That was actually the first problem.

I said, “Look, I play on mobile all the time. I’m on an airplane all the time, I don’t want to not be able to play when I’m offline, and only with a mobile device in my hand.”

And so the team’s first offer was, “Well, we’ll give you the auction house, or managing your home, or examining your character sheet.”

And I’m going, “No, no, I want to play the full game.”

So we had a huge argument.

Schultz There was lots of rolling of eyes?

Garriott There was lots of rolling of eyes. And “Richard is insane.” And it is still true that it may be an insane goal, but that is still my goal.

Which, as he explains, is actually possible given how they’ve set up Shroud of the Avatar’s client/server architecture.

Garriott [In a typical MMO, e]verything happens on the server. By definition, as soon as you say you want to play offline at all, that means the server functions have to be on-board with you on the client.

So that already doubled the footprint size, and it’s already big to begin with. So that’s why most people say “You’re crazy,” and write it off.

But once you accept that demand, that says the server-side has to operate on the client, what you would think of as the client side, immediately that opens up some other interesting opportunities. Which means that if you and me are playing together in an instance where only you and me are in it, we don’t need a server. One of us could be the server. (Leaving security issues aside for the minute.) And there are other issues that come up like, “What if I leave the instance, and you’re still in it. Now you need to be the master and take over operation of it.”

So there is some other complexity. But the advantage of saying that you don’t need a server, that if there’s a group of us in an instance that we amongst us can manage that instance, again, leaving security problems on the side for a second, that radically reduces our hardware need in the cloud if we can begin to leverage the assets within the playerbase.

Now, we’ve known for a while that Portalarium have been aiming for a very…decentralized server infrastructure for Shroud of the Avatar; even things like instances will be handled not on some central server in Amazon’s cloud or a closet in Austin, but rather will be shared between the players present in that instance, with most of the related processing split between their computers. What servers Portalarium will maintain, it is expected, will handle things like matching players together in the online play modes, and storing persistent world data.

But, as Garriott notes above, this decentralized approach also makes the idea of a mobile (or, more generally, offline) version of the game possible:

It allowed us—this moving the server-side into the client—meant that suddenly we had all these possibilities for the ways that you could play. You could play like an MMO, completely open, everybody in the whole world, we see them all on the same map, up to the limit of how many people we can put on a map. And then we make another instance, and then the people who don’t know each other, we find a way to slice people into one instance or another based on the friends graph.

It could also mean “I don’t want to see the friends I want to play with.” When I walk down the streets of New York City, 99.99 percent of the people I see walk by might as well have been NPCs, because I’m never going to have to see them again, I’m never going to have any interaction with them. And the same thing is true in a virtual world. Seeing people hop around the world in another MMO is not relevant. It’s only relevant when it’s someone I know, or someone I have repeat access to.

So it allows us to throttle back how multiplayer-feeling the game is, from completely-open MMO, all the way into soloplayer that gets it’s updates live from the world. And the nice thing is, even for those people playing soloplayer, they’re still going to see my shopkeeper, they’re still going to buy my trinkets. If they buy one of my swords, I’m still going to get the sale, I get the reward of the sale. It doesn’t matter that they’re in another instance of my home’s area.

Schultz So the NPCs are persistent across all instances.

Garriott NPCs are persistent across all instances.

Of course, the offline (and mobile, should it ever materalize) version of the game will be subject to a few limitations:

Let’s say you as a crafter made the Demon Sword of Doom, and there’s one of them. It’s very special, it has your name attached to it, it has five special gems in it, so it is easily perceivable as a more or less, one-of-a-kind item. Well then, we have two options as a developer. We either have to say that it’s not available offline because it is a unique item, and so if I’m offline and I go to your vendor, I don’t see that sword, because there’d be no way for you to sell it to me offline, and not end up selling it to someone else who is online. So we’ll either quit exposing those to the offline people, things that are unique, or, conversely, we’ll have to say, you know, the fact that somebody bought two of these in the universe is no big deal.

We’re probably going to err on the side of caution, which is to take all the limited edition digital objects and one-of-a-kind pieces and pull them out of the offline, period. But, conversely, if I was crafting offline, and I was making that Dragon Doom Sword to sell it to your vendor, it is held in that stasis until I log back on, at which point it pushes everything back into the metaverse.

Schultz and Garriott also go on to discuss the zone-based, dual-scale nature of Shroud of the Avatar’s world. Much of the reason for this design approach has to do with saving cost (building large-scale 3D open worlds is expensive), but Richard Garriott has been insistent about the inclusion of several features:

…there’s two zoom levels of the zones; there’s indoor maps and scenario maps. What we’re not doing is, when you go inside, you get a building interior map. I love as close to monoscale as we can get, but we need to go to at least two. And while there are lots of team members that have tried to talk me into making the interior of a house separate from walking around the perimeter of a house, because it would save the development team more pain, it was just too much of a reality-breaker for me. It makes it too obvious it is a game, versus a real place.

There’s actually quite a lot more to the interview than what I’ve quoted here; click on through to read the rest.

1 Response

  1. Sanctimonia says:

    Have to agree with Richard’s design decisions here. For the game he’s trying to make I think it will work amazingly well. My only fear is that the complexity will scale the number of bugs needing to be fixed, and we all know how that has gone in the past. Hopefully the use of Unity will limit the scope of bugs and maximize the resources (staff) available to fix them.