The pity that almost broke the game

The math I hadn't done

A pity at 1,000, on a base of 1 in 4,096, isn't a safety net. It's a metronome. The math is simple and I simply hadn't done it: (1 − 1/4,096) raised to 1,000 gives 0.783. That means 78% of the desvios any player was going to get came from the clock hitting a thousand, not from luck hitting 1 in 4,096. The rate I was publishing — 1 in 4,096 — had almost nothing to do with the rate the player actually experienced, which was 1 in 887.

That broke four things at once, and I only saw all four after finding the first one. The published rate was, technically, a lie: any dedicated player would work this out by dividing confirmations by observations, and the one promise that sets this game apart from its competitors — the math is public and it's true — would be dead before launch even happened. The calibration slider was disconnected from reality: doubling the base from 1/4,096 to 1/2,048 only moved the effective rate from 1/887 to 1/791 — an 11% improvement, when it should have doubled. I'd be calibrating a parameter that, in practice, barely did anything. Exploring the roster cost 4.6x more desvios than focusing a single species until the pity triggered — a game with 120 species whose own math told you to ignore 119 of them. And rarity was migrating onto the bingo card: seeing a desvio became a matter of scheduling, every two hours, and what separated players who had a lot of them was just the FRASCO ZERO — an item that, conveniently, sits behind the paid pass.

The fix

I raised the pity ceiling to 16,384 and the effective base-contribution variable (what we call k) to 4. The design rule I locked in alongside it: the pity ceiling always has to be at least four times the base's denominator. With this new ceiling, the fraction of desvios coming from pity drops to 1.83% — almost everyone, in practice, gets their desvio from the real base rate, not from the clock. The average converges on the statistical truth of the base: 8.94 hours at 450 encounters per hour. And only the tail grows — which is exactly the job a safety net is supposed to do, without lying about the average.

The second bug, found after I thought I'd already fixed everything

I published the new tail as thirty-six and a half hours. The whole thing looked settled. Then someone — actually, me, reviewing it for an unrelated reason — noticed a hidden assumption buried inside the formula itself: 16,384 divided by 450 encounters per hour gives exactly 36.41 hours. That checks out. But 450 encounters per hour of what? The math assumed, without ever stating it, that all 450 encounters per hour were of the same species you were hunting — single-species hunting.

That was never true in any real quadrant. A typical quadrant has around twenty different species circulating. With no filter at all, you encounter the species you actually want to hunt at a rate of roughly 22.5 per hour, not 450. Redo the math with the real number: the 16,384 ceiling takes about 728 hours — thirty days — not 36 hours. The odds I was about to publish were wrong by a factor of twenty, and nobody, myself included, had noticed, because the formula looked too airtight to check the assumption underneath it.

This wasn't a balance problem. It was the same core promise of the game dying again, from the inside, in a different way: publishing 36 hours and delivering 728 is false advertising, even with zero intent to deceive.

What closed both bugs at once

The fix was making mandatory a system that had already existed as a loose concept and had never actually been written: the Corrente. After clearing a quadrant, you pick one species from it and start encountering only that one — permanent while active, switchable off at any time, no progress lost when you swap. With Corrente on, the 450 encounters per hour are all the chosen species, and the published math finally matches the real experience.

The detail I found most satisfying, once I ran the numbers: Corrente doesn't change how many desvios you see. Across 16,384 encounters, with Corrente you expect about 4.07 desvios; without it, hunting freely, about 4.00. Practically identical — because the k=4 fix had already equalized focusing and exploring before I'd even thought of Corrente. What Corrente decides isn't quantity. It's which species. That's why it could become a free, permanent system, unlockable early, without touching any part of the economy.

The number I decided not to relitigate

The FRASCO ZERO — the item that guarantees containment of anything, no roll, no luck — speeds up the path to a desvio by exactly 50 times: about 447 hours on the standard path versus 8.94 hours with the item. That's pay-to-win, consciously accepted as a design decision, because the item is never sold in the shop — it only circulates between players, on the market, which gives players who don't pay a real route (save up Credit, buy from someone who has it) instead of no route at all.

What's still pending

The cost per containment and the real time to get a desvio both need to be recalculated from scratch under active Corrente, because it concentrates encounters in a way that changes both of those numbers — and I hadn't redone that as of the moment I'm writing this post. I'm not going to publish a number I know is outdated just to fill space. It stays marked as pending, with the counter-conversion formula already published for the day the recalibration actually happens, without wiping anyone's progress along the way.