Too Fast, Then Too Floaty: Tuning the Thrust
One constant decides how TUNL feels more than any other. I changed it twice in a single day after a player told me the game was too fast, and the second change taught me more than the first.
The control is not what most people think it is
In a Flappy-Bird-style game, a tap gives the bird an instant upward velocity. TUNL is different. Holding does not set your speed, it sets your acceleration. While you hold, an upward force is added on top of gravity, and while you release, only gravity acts. The ship carries its momentum through the switch, so if you let go at the top of a climb you keep rising for a moment. That is what gives the game its weight, and it is also what makes the numbers so sensitive.
Three constants define it: gravity pulls down, thrust pushes up while you hold, and terminal velocity caps the speed. The thing that matters most for feel is the pair of net accelerations: how hard you accelerate upward when holding (thrust minus gravity) against how hard you accelerate downward when releasing (just gravity). The ratio of the two is the game's personality.
How it got too fast
Over several passes I kept tightening the numbers, chasing a snappy, Flappy-Bird-like feel. Gravity went from 1,150 to 1,300, thrust from 2,400 to 3,400 and the terminal velocity from 820 to 1,080. Every step felt better while I was testing, because I was tuning to reduce the delay between my thumb and the ship. By the end, the net upward acceleration was about 1.6 times the net downward one.
Then a player on Reddit said the acceleration felt too fast, and that they were dying before the run's rush ever landed. That was the first unsolicited feedback the game had ever received, which is why it carried weight. The thing I had optimised, responsiveness, was the thing that made the game unforgiving.
Too soft, then somewhere in the middle
The first response was to walk thrust back to 2,700. That is much closer to gravity, with the net upward acceleration only a little bigger than the net downward one. The same day, I played it, and it was wrong in the opposite direction: floaty. The ship hung in the air, and climbing barely felt different from falling, which is a different way of losing the feel.
So the second change was a compromise, chosen by looking at the two ratios rather than by feel alone. The original snappy tuning was roughly 2,100 up against 1,300 down. The walked-back one was 1,400 up against 1,300 down. I settled on 3,100 thrust, which is 1,800 net up against 1,300 net down, roughly midway. Climbing still reads as more responsive than falling, but there is no longer a near-instant snap to the top speed. Gravity and terminal velocity did not move in any of it.
The lesson I took from it is that "too fast" and "too floaty" are two ends of one ratio, not two separate problems. Once I was looking at net-up against net-down, the target was obvious. Before that I was adjusting one number and re-reading it by feel.
The second half of the answer was the opening
Retuning the thrust fixed the feel but not the reason a new player dies quickly. When I measured a beginner's run, the median was about one second of flight. The launch animation alone was longer than that. And the pattern was clear: a beginner presses and holds, the ship climbs into the ceiling, and it is over. Nothing in the game ever taught the other half of the control scheme, letting go.
I ruled out a text hint. I had tried one, "HOLD to climb, RELEASE to fall", under the title and took it out the same day, because next to the button that already says to hold it read as redundant, and it crowded the row of buttons under it. The information was right. A sentence was the wrong way to deliver it.
So the opening coins teach it instead. Over the first stretch of cave, the coin line is forced onto a gentle arc that starts below the launch height and rises above it. To take the first coin, you have to let go and glide down. To take the second, you have to hold and climb. By the third you have felt both halves of the control, and nobody said a word. This worked with the rest of the change, a safe opening stretch with two hull scratches, which I describe in the pacing post.
What the tuning history is for
The reason I keep the record is that a constant like this drifts. It is one number in one file, it is easy to nudge, and every nudge is defensible on the day. I keep a history of every value it has held and why, so that the next time someone (including me) thinks the game would be better a little snappier, the answer is written down: it was, and players said it was too fast.
The input model is also fixed. Hold-to-thrust as an acceleration is the identity of the game, and the tuning is allowed to move around inside it but never to replace it with an instant impulse. Screens of different sizes get the same felt trajectory through a scaling rule I describe in the fairness post.
If you want to feel the result, the ship is one tap away.