Solve it at the limit

When building, you should set up a test bed to push things to the limit. Because if you solve things at the limit, you will find new exploit opportunities. Solving things at the limit gives you breathing room - and in that breathing room, you can consume trade offs. Solving things at the limit gives you budget. Budget that can be used to work around tradeoffs that others need to take.

This is why I chose arbitrarily large requirements for my online sand simulation game. If I can figure out how much state I can push, I can design around it. This allows me to create novelty

Going to the margins is to go to where few people have gone. Where no one thought to go, because it was too difficult, or required too much creativity. "Why do it that way? Why solve for that, specifically?"

Going where no one has gone is to have something that other people don't. It's how you get a moat. It's how you push the frontiers. By building things, and solving them, at the limit.

This is a trick I've learned from a very good engineer. When you think about things at the limit, you find exploits. I started doing this naturally, and before I knew it, I started engineering around things. Most people will think going to such extremes, is ridiculous. And most of the time, it is ridiculous. But, it's useful.

Thinking about things at their limit goes hand in hand with thinking about things according to their principles. E.G., what is a web browser? In principle, it's a turing machine. So, in principle, you can make all of adobe work right in the web browser natively (this turns out to be true!). Or; does a web browser have GPU access? Yes! Through WebGL! And what do WebGL shaders do in principle? SIMD. Which means you have a web browser that can do SIMD. Which means you can get a web browser to forward pass a neural network. At the limit!

The most important ingredient here is to be capable and optimistic that you can take things to such limits. Optimistic enough to try in the first place, and capable as to not fail.

The limit is the room you have available

You need to test it! You'll never be able to take all of the room available, if you don't try. And when you do try, you'll often find that you have quite a lot of room available. In that room, you'll find opportunities. Something new, something no one has tried before.

The limit is what can kill you

When thinking about system design or engineering, the first thing you ought to ask yourself is what are the both ends that this variable can possibly reach. Because it will reach it, and if you don't account for that possibility, your system will fail. This can range from the amount of relationships an entity can have, to the torque that a robot joint needs to output. You need to find the limits, and actually test them. See what happens!

Pushing things to the limits is useful. Things exist at the limits. Pay attention! To the limits!