Building Traffic on a Game Site
It seems as though the web ought to be a natural medium for
games; and many bold developers have bet on the assumption if you
it they will come. Sadly, unlike at the movies, it just ain't
so. The web is littered with game sites that ought to be
dead; and scores more have died and disappeared. Sure, if
are Yahoo.com (or name your favorite mega-portal) you can instantly
an overflow crowd for anything; but if you don't already have a
eyeballs, how do you get from zero to a viable community of players?
I'm going to present two cases, one sad and one happier, with which
I had some personal knowledge.
The Sad Tale of Midigames.com
My first example is Midigames.com. I followed this site
more or less from beginning to end because it hosted Lines of Action as
one of its many games. Dennis Rahaman, who built the site, did a
lot of things that were either obviously right or at least
His general plan was to build a great web site, attract players with a
free trial period, then eventually charge a small subscription fee.
Stage 1: Dennis quit his day job and devoted at least a year
to developing the software, which included a server, client, and LOTS
games. The games included many unique games, not available
each presumably with it's own small contingent of devoted fans. (Dennis
is a developer, not an inventor). The games included lots of card
games, chess variants, checkers variants, Lines of Action, and some
interesting hex games.
Stage 2: He launched midigames and contacted the obviously
parties (his is where I came in) who would help promote particular
He posted on rec.games.* and generally spread the word on the
Several hundred people downloaded his client and signed up for the web
site. On the way! Right? Wrong.
Stage 3: Dennis tried various things to populate his
but empty site. He tried mass mailings to his user base proposing
at particular times. He proposed tournaments with prizes. He
other game related mailing lists and spammed them. No response;
a poor response, not a weak response: zilch.
Stage 4: When the money (his savings) ran out, he got a day
and tried to market the site to companies that wanted content.
When last seen, his plan was to move to France and open a
The Happier Story of Tantrix.com
The original Tantrix.com was built "on spec" by a small software
with a plan much like Midigames. They saw Tantrix.com as a demo
their capabilities, from which they would launch a business building
sites for fun and profit. They built it, nobody came, they
to pursue more promising businesses.
When I first encountered Tantrix.com, it was in a state like
but worse; Not only was nobody there, but the site was
I volunteered to help fix it, and I've been "helping" ever
In the course of 6 years, traffic on the site has increased from
zero to approximately 40,000 games played per month. Here are the
things I think were the key elements to building traffic to the current
- Reliability. Unless players can play reliably,
come back very many times. This is a necessary but not sufficient
condition for a successful site. Midigames was always pretty
but it didn't save them. One of the things I added to tantrix.com is an
extensive error logging system which has exposed innumerable
- and knowing a problem exists is a necessary first step toward fixing
it. It should be noted that lots of strange things happen "out
especially with various versions of java on various browsers to contend
with, but also due to network glitches and unexpected sequences of
- Robot player. I originally thought the robot would
to keep players in the site until other players arrived. That has
proved to be true, but to my surprise, lots of players spend lots of
playing the robot, even when other humans are available.
something like 3/4 of all games played are human-vs-robot games.
I think it's important that the humans playing robots are mixed with
humans playing other humans, so they have the opportunity to interact,
and so the robot players are part of the pool of potential human
- Ranking system. People love to keep score, and
over the long run keeps them interested in coming back for more.
It's more important for the ranking system to give players the
of change than it is for it to be completely fair and accurate.
has a lot of different score related functions.
- Tournaments. Self-scheduled
tournaments (where the matches are played at times negotiated between
the players) are very popular. Site usage always takes a jump when a
tournament is on. The key thing is that someone has to organize
- Fanatic/site tending players. The most important
get players to a site is that they find other players already
The best way to do that is to have a few fanatics hanging around, ready
to pounce on newcomers and make them welcome. Under some
assumptions, you need a pool of 200 casual players to reach the state
finding an human opponent is reasonably likely at any time. Three
fanatics who spend 8 hrs/day in the site would do the same.
- Design to promote interaction. You should have
areas, and all players in the site should remain in contact with each
the whole time. Even game nerds like to interact with each other.
- Low barriers to entry. Java is better than a
client, because it is always there. Guest accounts will
the curious who would be put off by any registration procedure.
my surprise, guests have played more games on tantrix.com than any
It should also be noted that that a lot of the first time visitors
to the site following a link from an physical game they bought, not a
from someplace else on the web; and that the site has had little direct
effect on the worldwide spread of Tantrix. If there were
watching this, they would be shaking their heads in dismay.
However, the web site has been important in another way. Fanatic
online players have gone on to become distributors responsible for the
worldwide growth of Tantrix sales. The web site found and
nurtured the fanatics who make the game grow.
The Next Phase:
The next phase (starting March
2004) of the saga is to
see if the relative success of Tantrix.com can be
replicated. Boardspace.net was created by recycling most of the
code and know-how from Tantrix.com, as well as applying the lessons
learned. Boardspace.net is to be a multi-game site, not
affiliated with any particular game manufacturer. First
game is Zertz, which is in a similar
position to Tantrix in 1998; it is a reasonably well known and
game with a live inventor, but not one that is hugely successful.
Following Zertz in pretty quick succession over the next 2 years were
Lines of Action, Plateau, Yinsh, Hex, Trax, Punct and
Gobblet. I had specific motives for each game, all aimed at the
"brass ring" of achieving a self sustaining community. It's
not there yet. Here are some of the ideas behind the
choices and thoughts about the results.
- Have insanely great games. Once
people are exposed to them, players will be addicted and bring their
friends. This was the original motivation for Zertz.
I thought (and think) that Zertz is a really exceptional game.
Maybe you agree, but there hasn't been a spontaneous explosion of Zertz
players at boardspace or anywhere else. Gobblet is another
application of this theory - it's lots of fun and very simple to
learn. We'll see.
- Get other people to work for you.
Not in the literal sense, but for example, who should be more motivated
than an inventor with a game to sell? Zertz, Yinsh, Punct, Trax,
Plateau and Gobblet all have live inventors and real distributors
trying to sell the games. They have all been pleased to see their
games at Boardspace, none has made it prominent part of their marketing
or promotion effort.
- Share the wealth. At
least 90% of the effort to get a web site with one game is
unrelated to the game itself. Boardspace has all that machinery,
and a reasonably good framework to add new games, so why not make it
available? I added Hex to boardspace as a "programmer's sample"
of the simplest possible game. Several game inventors had
expressed interest. So far none has actually followed through.