The premise

Breaking Iron runs structured, filmable gaming challenges. Every episode is the same shape: a hypothesis, a fixed rule set, a fixed sample of matches, real tracked data, and an honest recap of whether it worked.

Why the rules are fixed

A challenge with rules you can adjust halfway through isn't an experiment, it's a highlight reel. The rule set is decided before the first game and doesn't move, including when one of the rules turns out to be annoying. There are never more than five, because past five nobody can hold them in their head — me included.

Why the sample is fixed

Ten games, decided up front. Stopping at the point the numbers look good is the easiest way to lie with data, and it's the one that's hardest to spot from the outside. The sample size is set before filming and the run finishes whether or not it's going well.

Why the data is entered by hand

Every match is read off the post-game scoreboard and typed in during the session. An API would be faster. It would also mean never looking at the numbers, and looking at the numbers is most of the value — you notice things in a scoreboard you'd never notice in a spreadsheet someone else filled in.

Why the failures stay up

Some of these experiments don't work. The vision one where the number went down. The idea that seemed obviously correct and simply wasn't. Those episodes get the same treatment as the ones that worked: the full tracking log, the chart, and a takeaway that says plainly that it didn't pan out.

A channel that only ever confirms its own hypotheses isn't running experiments. It's making content that happens to have numbers on it. The difference is the entire point.

The name

Iron is the bottom rank — in League, in Valorant, in most games that rank you at all. It's also what you're holding when you start working steel. Steel that's being worked changes colour as it tempers: straw, bronze, purple, blue. That sequence runs through this site, marking how far along an experiment is. It seemed more honest than a progress bar.