Principles

Most of what is wrong in this industry runs on people not knowing how it works. You cannot fight that with an argument. You fight it by handing people the thing itself, so they can look inside and see for themselves.

That only works if it is free. The moment there is a price on it, the people who most need it are the ones who do not get it, and the gap you were trying to close gets wider instead.

So I open source it. All of it. Open source the tools, open source the research, open source the world.

I will never sell that.

This costs me money. The lab runs on my paycheck and the tools I write get used by people who will never know my name. That is fine. It is the entire point.

The most useful thing I ever learned is that there is no wall. I kept expecting to hit a subject that was genuinely past me, and it never happened. That surprised me, and it still does.

Look up what the word actually means. Complicated is a lot of moving parts. Sometimes it is nuance. That is the whole definition. It says nothing about whether you can follow it.

So when something looks impenetrable, it is usually one of two things. Either nobody has bothered to explain it in order, or there are more parts than you can hold at once. Both are fixable. Write the parts down and take them one at a time.

The feeling of being out of your depth is real. It is just not evidence.

I watched people rebuild the same modding tool from scratch, over and over. Each one got about halfway, decided it was not good enough, and deleted the repo.

Every one of those deletions threw away the only useful thing that had happened: the part that already worked. The next person started from zero again, hit the same wall at the same place, and deleted theirs too.

Leave it up. Even unfinished. Especially unfinished.

Someone else can build on a half-built thing. Nobody can build on a deleted one. The embarrassment you are avoiding is smaller than the work you are destroying, and it is your embarrassment against their month.

If it truly is abandoned, say so in the readme and leave it anyway.

My instinct used to be to hide it. Either talk around the problem, or act like it was not happening while I quietly searched for a way out.

That never once worked. It just moved the cost. Hours went into looking busy instead of into the actual question, and the question was still there afterward.

Being honest about it feels better. That is the part I did not expect. Not more virtuous, better. The moment you say "I do not know why this is broken," the problem becomes something two people can look at instead of something you are carrying alone.

It also gets you an answer faster, which is the boring practical reason to do it.

There is a split people accept without arguing about it. Either you build something genuinely hard, or you are good at presenting things. Depth or showmanship, pick one.

I do not see why. "I made something complex and unique" and "cool, what can you do with it" are both fair. The second is not a lesser question. It is the question everyone who did not build it is going to ask.

Charisma in someone who actually knows the material is rare, and that scarcity is the whole reason it works. Being able to explain your own thing is not a betrayal of the thing.

Do both. The work has to hold up when someone reads the source, and it has to be worth their attention before they get that far.

This site used to say "8+ live services" and "99.9% average uptime." Nothing measured either number. They were typed in because the layout had a slot shaped like a statistic.

That is not a rounding error, it is a lie with a percent sign on it. And it is the easiest kind to tell, because a number reads as evidence whether or not any evidence exists.

The rule here now is narrow and absolute. A figure appears on this site only if something measured it, something configured it, or it is a plain fact about my own history. If the source is unavailable, the number is not shown. Not estimated, not held over from last time, not replaced with a friendly approximation. Removed.

You will see gaps because of this. The live status page drops tiles when it cannot reach what it monitors, rather than filling them in. A gap is information. An invented number is not.

There are two tests in this repository that fail the build if an unsourced figure comes back. I needed them, which is the honest part of this.