XP is a lie I tell on purpose.

I want to be honest about something that sits a little uncomfortably behind the rank shields and tier names on this platform.

The XP formula is made up.

Not entirely — it's grounded in something real, which I'll get to. But the specific numbers? The exact amount of XP a hard problem is worth versus a medium one? I estimated those. I could change them tomorrow by 20% in either direction and the system would still function. The tier thresholds are the same story. Wood III to Bronze is a particular number that I chose, revised once, and then left alone because it felt roughly right. "Felt roughly right" is not a rigorous methodology.

So why does it work?

What XP actually measures

The formula has two inputs: problem difficulty, and a per-problem override I can set manually as the coach.

Difficulty comes from the upstream catalog the platform is seeded from — a rough 0–4 scale. Harder problems are worth more XP. The override knob lets me adjust individual problems that the difficulty rating doesn't capture well; a problem rated "medium" that I know trips up advanced students for subtle reasons might get a bump. One that looks hard but has a clean trick might get trimmed.

So the formula gestures at difficulty, and I tune it toward my actual assessment of the work required. That's still not a clean measurement. But here's the thing: what the XP number ends up tracking isn't "how skilled are you" — it's "how much coach-validated effort have you put in over time."

You cannot earn XP by gaming the system. A submission goes through me. I read it, I check the logic, and I either approve it or send it back. Approved submissions earn XP. That filtering step is what makes the number meaningful, even if the number itself is arbitrary. The unit is invented; the thing being measured is real.

XP is a receipt for work that someone honest has looked at and signed off on. The amount on the receipt is a guess. The fact that the receipt exists is not.

What the rank actually says

I want to be clear about what a rank on this platform is not. It's not a Codeforces rating. It doesn't tell you where a student would place in an open competition. Two students at the same tier might have very different raw competitive ability — one might be a methodical worker who grinds through problems slowly and thoroughly, another might be fast but selective. The rank captures something about the former; it says less about the latter.

What it does say, with some confidence, is: this student has put in the reps, and someone was paying attention when they did.

That's a real signal. A Gold-tier student on this platform has had a meaningful number of solutions reviewed and approved by a coach who knows what they're looking at. That's more informative than a raw solved-count anywhere, because solved-counts don't include a quality filter.

I'm also not going to pretend the dopamine of leveling up doesn't matter. It does. Reaching a new tier feels good, and that feeling is a legitimate part of why the system functions. The problem with olympiad training is that the middle stretch — after the initial excitement, before you're good enough to be visibly improving at competitions — is genuinely boring and hard. You're grinding through problems that mostly feel out of reach. XP and tiers scaffold that middle distance. They give a student something to move toward on weeks when the problems feel opaque and the competitions feel far away.

I made peace with this. The gamification is real and it's intentional. I just think it's worth being honest that the scaffolding is scaffolding, not the building.

When the system works

The tell that XP is doing its job is when it stops being interesting to the student.

Early on, a new student will check their XP after every approved submission. They'll know exactly how far they are from the next tier. The rank shield on their profile is something they care about visibly.

Later — and this doesn't happen with every student, but when it does it's unmistakable — the XP stops being the point. The problem is the point. They submit because they want to know if their solution is actually correct, not because they want the number to go up. They'll ask me about the problem after I approve it, not about how much XP it gave them.

At that stage, the XP system has done everything it was supposed to do and become invisible. The arbitrary number served its purpose: it kept someone engaged through the hard middle until the work itself became the reward.

That's the measure I actually care about. Not the tier, not the leaderboard position — whether the student eventually stopped needing the scaffolding to show up.