Developers Diary
             The making of the world's greatest
    Computer Role Playing Game (RPG) (fingers crossed!)

    The story so far - by Graham Raven (Game Developer).
 (Also with bits by Michael Raven, the actual monkey who does
                     the programming!)

In the beginning. Pre-1999 I played 'Mordor' (RPG) a lot (i.e.
for three years! It's supposed to last 200 hours!) Suddenly around
4am one morning it dawns on me that I'm seriously into RPG's and
hey, why don't I create the 'best' RPG in the world ... as
according to my own definition of 'best' of course. (yes, I was
playing at the time!)

Problem: I can't program. Solution: Get someone else to do that
bit for me!  (Can you guess who? Go on, can ya? - Michael)

Following two years. Plan, imagine, discus, play RPGs and steal
the best ideas (this is of course is very much an 'on going'
exercise!) The discussions are still very much a highly favoured
pastime and frequently take place at a village pub whilst we check
up on the quality of the local ale.

2000AD. Actually start work on the world's best RPG!

Basic aims. Version one of the game should be 2D tile based - we
can worry about full 3D later (confident or what!?). Fellow game
consultant and programmer (My son Mike) agreed from the start that
'Conversation' is sadly lacking in current RPGs. A unanimous vote
agreed on the prioritisation of the Conversation Engine and
another round of drinks was ordered!  Even though we couldn't
initially do exactly what we wanted with a database, we were able
to plan. Little by little we would overcome the problems. It had
to be a database of course, it was the only way to handle the mass
of information we wanted to work with (three to five hundred
characters all with their own story to tell!) Also we established
the excellent strategy of beg, borrow and stealing whatever ideas,
resources etc that we could in order to smooth the progress of the
game. There's a lot of free stuff out there on the net, and
surprisingly there's a lot of people who are happy to help you
with your project - you just have to hunt them down.

A few of you (like Sue for one) already know of my epic RPG
aspirations that started many years ago. I've always played RPGs a
lot, they are my favourite game (they are the ONLY game!), and yet
no matter which RPG I played I always had feelings about "Well,
that bit's wrong!" or "I could do that better!" Initially what
were vague thoughts and attitudes about 'how such and such a game
ought to be run', gradually crystallised into our current game
project. As Mike's programming knowledge increased year by year
(he's doing a graduate degree in business IT), it became
increasingly apparent that we actually could do the seemingly
impossible - we could really make our own game!

Now a great many people 'out there' would love to make their own
RPG. Many try, and many fail. I'm happy to say that so far we
haven't failed yet and after a good three years are still enjoying
ourselves enormously, and aren't half-way through all the local
pubs yet which we need to evaluate as potential 'meetings' venues.
I'm quietly confident that we've got the approach just about right
for this project, enjoy ourselves right from the start and if
we're still working at it in ten years time well we'll still be
enjoying ourselves! By the by, we are occasionally asked, "Who are
you making this game for?" The answer is easy, "For us!" When
eventually the game is up and running it will be put on the net,
free for anyone who cares to download it, but it only has to
satisfy us!

As I often admit, my own programming skills are virtually non-
existent and when we were starting out Mike was sore put out that
I couldn't understand even elementary programming, and so he
determined to teach me the basics. With some effort I grasped the
idea of pseudo code - spiting out in words exactly what it is you
want the program to do - step by step - display such and such an
image - run fight program when a bad guy appears - show picture of
peaceful sky when you are dead, play music, display text message
"Oh Dear!" - etc. (One day he will attend and complete a course in
Visual Basic! - Mike)

Amazingly enough with Mike's persistence, I was forced to grasp
the strange relationship between the meaningless hieroglyphics
which are Visual Basic code and the simple map editor which we had
created. I was so pleased when it worked!  Basically the editor
made a grid which you could move about on and place numbers in the
different locations. Mike said that if we could place numbers in
different locations then we could do the same thing with graphics
such as bitmap tiles.  (representing grass, water, woodland etc)
Suddenly we could make a map! 'Numbers' became squares of colour
and squares of colour evolved into textures and images of the
landscape.

Now at a pinch I can write pseudo code, and with extreme badgering
more or less follow some simple VB code - this represent a huge
advance!

So I largely rely on Mike to do the clever stuff as far as the
program is concerned and I have embraced the official position of
'Game Developer' in our team of two. I was later to learn that
this 'job title' not only embraces thinking beautiful thoughts and
play-testing games, but also making hundreds of graphic tiles and
writing the game's huge gossip files plus the conversation
elements of every NPC in the game! I also tend to sample more ale
than Mike at our planning sessions; it's a tough job but someone
has to do it!

Of course the simple map editor was improved enormously over many
months, and inevitably we 'improved it' one day to the point where
it didn't work at all! It was bound to happen - This is an
occupational hazard I'm told!

Thankfully on my surfing the net some six months ago I came across
the Harry Riggs project; people who are very much into the kind of
thing that Mike and I want to do, only these are professional
software consultants! (Brains the size of Southampton!) Regardless
of our differing abilities we share the same passion for game
development and have often shared ideas with each other. The only
difference between their efforts and ours of course is that they
are simply brilliant.  Having looked at our code for five minutes
my friend at 'Harry Riggs' said "Ah, just change that bit!" and
after that everything has worked fine.

As I've already said, one thing we have always agreed upon is that
conversation ought to be at the very heart of our game.  We knew
that we couldn't make state of the art graphics, nor did we really
want to. Characters can become very convincing and believable even
in a storybook, so graphics aren't always essential to bring a
'world' to life, but good conversations and stories and characters
are! I find it irritating that modern 'state of the art games' are
still being made that have NPCs with idiot conversations (no names
mentioned) Whilst it is very difficult to have truly convincing
conversations with a computer program (i.e. so good that you
aren't sure whether it's a machine or another human that you're
talking to!) we still feel that this is one area of game play that
can be improved upon an awful lot.  We also believe that it's not
a question of very clever programming being needed, but simply
effort and 'sheer grind' on the programmer's part. If there was
one problem we aimed to crack, this was it!

So as I often do, I searched the net, this time looking for
'conversation techniques' and lo and behold up came a number of
articles at the Gamedev site - exactly what I had been looking
for! After a lot of research and several pints in various
locations we plumped for the old tried and tested 'keyword and
response' technique. A crude prototype database was created and it
nearly worked at the first attempt!  (it's scary when that
happens!) You enter a keyword and the NPC responds with a given
response. If you 'befriend' the NPC (by basically talking to him a
number of times) she/he takes a liking to you and starts to tell
you more.

i.e. You meet Fred and say hello. Fred recognises 'hello' as a
keyword and makes 'his' chosen response. What that response might
be depends of what sort of character Fred is.  He might say
"Hello", he might run away, he might attack you on sight (it takes
all sorts to make a world!) So far, so good.  We knew this
technique has its limitations but we believed that it could be
developed a lot further. The conversation engine was based on an
Access database and as usual, we gained help and expertise
anywhere we could. It's surprising how many people will help you
if you ask them nicely! One good friend by the name of Yvonne knew
not a fig about programming but did know the practical side of
using/setting up databases inside out - linking tables and all
sorts of weird stuff like that; a very useful friend to know in
the early days!

The original conversation engine has progressed in fits and starts
for over two years now and currently manages several hundred NPCs
all of whom have multiple Friendship Ratings (FRs), full natural
statistics and alignment plus the possibility of Artificial
Emotion which is generated from the stats and alignment. We still
need unique images for all the characters, but we'll worry about
that at a later date. The general ideas are something like this -

You chat to a game character and assuming they don't attack you on
sight, they will gradually get to know you a bit. Little by little
your FR will improve with that character (though to what extent,
you won't know until you try!) and they will be willing to tell
you a little more than they did when they first met you. Will they
have anything of interest or value to tell you? Who knows? You can
only find out by getting to know them. FR is improved in a number
of ways; simply talking to a character often will invariably
improve your friendship with them, to some extent at least.
Sharing a drink or several with them may help - depending on the
character - as may giving them gold.  Performing some heroic act
is also pretty good at making friends, though it doesn't always
work!

At first they will tell you very little, but as you befriend them
increasingly, you get to learn more and more information. Not
everyone has great secrets to share with you, but you'll have to
be their friend in order to find out!

Then we got thinking about just what information an NPC would have
to share with the player should they meet in the game world.
Obviously their name and background, their job and such personal
information as 'likes and dislikes'. Hmm, and then there's
'Gossip'. In this context 'gossip' covers news of any kind, and
all gossip will be stored in the games 'massive' gossip files.
Gossip will include events and occurrences of all kinds, from
outbreaks of plague and earthquakes (very rare) to drunken fights
in the local pub (very common!) Even some characters dreams will
occasionally turn up in the gossip files. All events in the guilds
such as who's reached a certain rank this week (and who failed!)
are also 'gossip' (and who'd been robbed, lost their cat, fallen
ill - you name it!). We dislike the notion of predetermination as
far as future events are concerned, so there will be no fixed date
for such things as outbreaks of plague, rather just a low chance
of them occurring. Some things will of course, always happen on a
regular basis and will follow a script eg. there will always be a
harvest at about the same time of year, though whether it will be
a good one or not again is subject to chance (and the weather).

In order to try and give the NPCs more individual conversations we
aimed to use their natural stats, guild, race and alignment to
govern which parts of the gossip file a given character can
access. Thus an Elf will know more about elves in general than say
an Ogre, and a Warrior will know far more about 'Warriors' Guild
gossip' than a Sorcerer would. A barmaid, if she's the friendly
type, will know a great deal about local gossip though little
concerning scholarly matters - for the latter, find yourself a
scholar (scholars are very knowledgeable about certain subjects,
such as puzzles for instance, though often sadly ignorant
concerning common gossip).

We also plan for different races to use different dialects in
order to build an impression of 'character'. Ogres are more likely
to say "I'm sayin nuffin" where as an elf might reply "one's
thoughts are one's own" This will involve a lot of effort in
writing characters responses but could work very well.

We have yet to implement all of these features into the
conversation engine, but are confident that we can do it.

I must say I'm persistently envious though that the Harry Riggs
NPCs can exchange information with other NPCs in their game -
something I haven't quite got my head round yet, but I'm sure we
will - it's just another aspect of which NPC can access which part
of the gossip file.

(Hmm, mental note - whatever the player 'does' - change guild
rank, who he/she fights, hmm other things - must also go into the
gossip files! This needs some work!)

Then we thought about movement and NPCs. I have played games where
the characters apparently move about entirely at random, or
occasionally stay in the same location all the time.  Hey we're
trying to simulate reality here (well, sort of!) Most NPCs will
tend to stay in the same village, town, dungeon etc though move
about within their area - they have lives to lead after all, so
that includes going home to bed, eating etc. We aim to make
certain that the barmaid ISN'T behind the bar in the middle of the
night! So Mike devised a system of 'drives' for the NPCs. A
character will want to feed, sleep, and work (usually!) There are
specific locations associated with sleeping and working, possibly
feeding too, for each NPC, and the characters will follow general
routes to their place of work (place of rest etc). Most of the
population will be pretty darned regular in their habits, as if
following some invisible internal clock, but we think this is
still a better model that having them just stay in one place. This
programming needs some development if we're going to avoid
characters falling asleep half way home or eating whilst they are
asleep, but we are halfway there.

A few characters will be 'wanderers' and will turn up all over the
place. I suspect that they will have individual coding to govern
their movements.

Meanwhile by way of a break, we have concentrated on getting the
map up and running - we thought that being able to travel across
the landscape might inspire us in other areas of the project. So
far over six hundred 32x32 pixel tiles have been created to date,
covering most aspects of landscape. We still need to work on tiles
for 'indoors', dungeons, etc. We've often played games where there
simply wasn't enough variety of tiles used to generate a decent
landscape - fingers crossed we've managed to do a half-decent job
here! (Ahem, I just have to tell you all about the size 32 x 32 -
this is the third size we've had so far! First 37x37, then 16x16,
now 32x32.  Hopefully this will be the last change! - Mike)

This is the current state of our project. Another hundred or so
tiles (I hope it's only a hundred or so) and I can start placing
characters in various locations. That should help to inspire new
(potential) scenarios and events, plus we'll probably need another
hundred or several characters sketching (not to mention visiting
'The Monk's Walk' in a nearby town, and 'The Gnu' in North Newbald
looks very promising) . any budding artists out there who'd be
interested in helping out? We're looking at straight forward
monochrome line drawings!

The great work continues ...

C.G.RAVEN - Game Developer (Quality control beer sampler, etc)

Places to visit:
www.gamedev.net - for all your programming needs
www.planetsourcecode.com - when you can't think of how to do it
http://www.theharryriggsproject.inuk.com/ - to see how to do it
properly!
                              - o -