Today, I started building North Is Not Up around one rule: the game should never tell the player where they are.
There would be no blue dot, minimap, route arrow, suggested path or distance counter. If I wanted the game to feel like orienteering, the player had to read the landscape, orient the map and decide when to trust a bearing.
That meant the first job was not building a huge course. It was proving that a map and compass could carry the whole experience without a conventional navigation interface quietly doing the difficult part for the player.
Making the map independent
The map and player position were deliberately separated from the start. Opening the map could not centre the paper on the character or reveal where they were standing. The sheet only knew the world geography and the direction it had been rotated.
I used a fixed coordinate convention for everything: east is positive X, north is negative Z and bearings increase clockwise from north. It sounds like a small implementation detail, but it became the contract between the 3D world, the topographic map and every control location.
If a bridge, path or boulder moved in the landscape, its mapped version had to move as well. An approximate decorative map would have made every later navigation mistake ambiguous: did the player misread the terrain, or was the game wrong?
The compass had to behave like a tool rather than a heading display. The player could position it on the map, aim the direction-of-travel arrow and use the bezel to relate a chosen route to grid north. Back in the world, the physical needle provided the familiar “red in the shed” check without turning the answer into a number on screen.
Building the Woodland Training Area
I kept the first playable world to 100 by 100 metres. The Woodland Training Area was large enough to contain paths, terrain and recognisable features, but small enough that I could test the navigation loop without spending hours constructing scenery.
The course taught the essential sequence: open the map, understand its orientation, configure the compass, close the map and follow the bearing in the landscape. The character began with a random facing direction so the first task could not be solved by assuming the camera already pointed north.
The control itself had no glow, floating label or distant marker. The player needed to follow the line and identify the feature. That was an important early test because it showed whether the central idea remained understandable once the usual game assistance disappeared.
The first version was still visibly a prototype, but it proved the rule worked. I could navigate using the same information I would expect from a paper map and baseplate compass, and the game did not need to reveal my position to make the interaction usable.
Expanding into Hawthorne Ridge
Once the small training ground worked, I began a separate 400 by 400-metre course called Hawthorne Ridge.
This was the first attempt to see whether the navigation held up across a landscape rather than one short lesson. I placed a physical start, six controls that had to be visited in order and a finish. The timer and split system would come next, but the complete sequential route was already in place by the end of the day.
The larger environment immediately exposed a different class of problem. In a tutorial, one path and one landmark can carry most of the scene. Hawthorne Ridge needed open meadow, woodland, slopes, a river and enough variation for six legs to feel distinct.
I went through several passes on the daylight, terrain and natural environment. The Alder Beck stream was particularly awkward. Its depth, banks and water surface all needed repeated correction so it looked like part of the land rather than a trench cut through it. A crossing also had to be visually readable without becoming an oversized signpost.
What the first day established
By the end of today, the project had a name, a permanent training area and the structure of its first full course.
More importantly, I had established the rules I would keep using: the map must match the world, the compass must remain physical, and the player must make the navigation decisions.
The game was not yet a finished race. Hawthorne Ridge had no complete timing and results layer, the tutorial still needed clearer guidance and the shared menus and settings did not exist. Those gaps were useful because they made the next stage obvious: turn a working navigation experiment into a complete game structure without weakening the original idea.
