We wanted a huge world at launch.  We knew we were going up against other MMOs that were already out there, several with expansions already released.  Also, with our goals of a seamless world and a view that went on for miles we knew the size of the world would be key.  We also wanted to lay out the majority of the world, including what would be released as expansions. 

The game designers used map making software and produced this huge world.  It was crazy big, although it was hard to tell exactly how big because we were just looking at a map.  We hadn’t tried to create any of the world in-game yet.  The art team needed time as did the graphics engine (taking the Unreal 2.5 engine and making it seamless world capable took some time).  We then decided on three major continents, Thestra, Qalia, and Kojan, and then what races would start out where. 

We knew that those three large land masses would accommodate a lot of people.  Travel time could be an issue, so we made sure we’d offer vehicles at a relatively low level (we used the term vehicles loosely and to include horses, ships, etc.).  On message boards and in the FAQ I let people know about the size and, although we were not fans of teleporters, I posted that they may be needed to some extent and that we’d determine that in beta testing.

This led to two major errors.

First, when it became time for the artists to start building these land masses, it took longer than was expected and hoped.  The easy solution at that point would simply be to shrink the world.  But when this problem occurred to us, when Thestra was already mostly done and Qalia in progress, it wasn’t an easy fix.  People were saying ‘let’s shrink Kojan’ or ‘let’s just cut Kojan completely’.  But shrinking it too much would make it much smaller vs. the already built Thestra and Qalia.  Cutting it completely would mean we didn’t have a home for the races that started there.   So we ended up cutting parts of it away, and the game launched with Kojan being smaller than the other two major continents.  

Then, the art team went back and polished Thestra.  The tools were a lot better at that point, and the artists better versed on how to build a good looking world.  And even with re-visiting Thestra, if you really look, I think Kojan is the best looking continent by far.  But the bottom line was that we spent too much time building three continents and making them look good.  We should have had fall-back positions, enabling us to make a smaller world with more starting areas in fewer continents.

The second major error had to do with under-population.

It’s hard, perhaps impossible, to launch with exactly the right amount of content and world size.  With EverQuest the problem the majority of the time was over-population.  There was only a certain amount of content, of dungeons, of outdoor areas.  When the game took off being a much bigger hit than we had assumed it would be, over population was the big issue.  Too many people per server meant too much fighting over limited resources.  Players grew frustrated.

The fix for EverQuest, though, wasn’t all that hard to implement.  We came up with an upper limit in terms of server population during peak hours.  If a world server was exceeding that limit, we’d launch another world server and then encourage people to migrate.  We also would split servers if necessary.

What happened with Vanguard, however, was very different.  Under-population is much worse than over-population.  As mentioned, if a world server was over-populated, you can add another server and/or split a server into two worlds.  And until you did so, people would complain (and rightfully so) that the world was two crowded and the fight over limited resources (zones, items, etc.) would grow too intense.  That said, players usually would not quit the game.

But under-population creates an empty world in terms of other players.  A big part of Vanguard (and EQ, for that matter) was about grouping.  If we were going to push grouping, then people needed to run into other players.  They could then group, get to know each other, and feel part of the world.  Obviously, there was a certain population density needed for this to reliably occur.  And, even with the game offering soloable content, players still want to feel like they are part of a living world.  They may prefer not to group, but they still want to see other players, duel them, trade with them, etc.  And, unlike the case of over-population, people would indeed quit because of under-population.

So if we knew because of our bigger world that we would need to support a larger number of players per world, what went wrong?  Server performance.  With EverQuest the world server would become over-populated in terms of resources and crowding before the actual server would become overloaded.  I think this gave us a false sense of security and when, during beta, we discovered that the software and hardware limited us to too few per world, we really didn’t know what to do.

So we launched with a huge world and unparalleled in-game views of that world, but with hardly any other players around with whom to share that world.  Adding teleporters helped, but not enough.  Community, which is an important glue when it comes to MMOs, didn’t build because of under-population.  The world seemed empty, it was hard to find groups, etc.

Looking back I think it would have been a lot better to make the world more modular.  The design, the content, and the racial starting areas should have been laid out such that we could shrink the world without it interfering with our plans.  And then we should have created test areas much earlier on that would have allowed us to artificially populate a region with lots of players.  This would have revealed a population cap lower than what we had assumed.  Then we could have reduced the world size accordingly.  I think the fact that we correctly and relatively easily dealt with EQ’s over-population lured us into a false sense of security.  We were simply so ambitious that we were blind to a lot of this, and by the time we realized what these problems were it was simply to late to go back and conduct a major overhaul, especially given that we were short on development time anyway.