I spun this post out of RC Weeknotes 03.


“Make something people want” is the cardinal Paul Graham maxim. It’s at the heart of YC lore.

I heard it. I understood it. And I also knew that I hadn’t fully assimilated it yet.

But after my experiences making some games these past few weeks, I think I finally get it. And I’m updating massively in favour of interactive experiences that afford fun.

As an ML Engineer, I’ve built useful and important and performant things. But it’s been many years since I built something that my girlfriend wanted to show to her parents. Yet when I made a game that runs in any browser, it’s as though she couldn’t help but share it.

Normally, it’s an exercise in futility trying to explain this stuff to anyone who doesn’t understand both Shannon Entropy and how to derive Backprop from scratch. But just about anyone can immediately understand and engage with my silly little hedgehog game.

Several people at RC were still posting highscores and comments about the game multiple days after I first shared it. And I received the ultimate endorsement in the world of games dev:

Message about his 7-year-old playing my game

Two aspects of this are noteworthy:

Fun is honest feedback

Firstly, it’s tremendously hard to know if a game concept will be fun in advance. You only really have an inkling once you’ve built an MVP and tried it. And you only truly know once you’ve seen whether other people come back.

In this regard, there’s a similarity between comedy and games and products. You don’t know if the joke is funny until you try it and get a laugh. You don’t know if the game is fun until someone’s kid is hooked on it. You don’t know if you have a product someone will pay for until it’s in the market and they’ve put their first payment through. And just like with all those endeavours, you mostly fail. It’s a minority of jokes that are funny and games that are fun and products that are worth paying for.

I’ve treated the “serious” code I’ve written and the products I’ve shipped as a class of their own. But going forward, I’ll think about them more like games and jokes. I’ll just try a bunch of stuff and most of it will suck, but occasionally I’ll “kill” and it’s that work which endures. It’s also much more fun.

Fun is flow

The feedback loops on small solo games are obvious and delightful. The hedgehog either feels good to move or it doesn’t. The hoop is either the right size or it isn’t. You can rapidly assess, tweak it, and iterate.

When the feedback loops are tight, you surf the flow state all the way to something you’re proud of and other people want.

This is much harder to achieve when you have longer cycles and less tangible interactions. Playing games and making your own is fun and rewarding. It’s work you have to force yourself not to do. Pay attention to how that feels and seek it out in whatever work you do.

But seeking it out isn’t enough. You also have to engineer it in.

The game that runs in a browser has lower friction than the one that you have to install. The one you have to install has lower friction than the one you have to build a custom rig to play.

Lower the friction to using the things you make to get the earliest and clearest signals. Then invest in the things that you and others keep coming back to.

“Make things people want” is tough advice to follow without a hunch of where to start. So begin by making games people play and we’ll figure it out from there.