I don't actually know how I learnt this or how it came about, but for as long as I can remember, I've always thought this way about a lot of things.
Maybe it goes back to childhood. I wasn't a doer — I was more the quiet observer, watching how things worked and functioned before I ever tried them myself. Somewhere in that habit of watching, a kind of backwards thinking took root.
It didn't have a name for a long time. It just became more apparent, and more useful, once I landed in VFX. But it also has a limit, and I didn't understand where that limit was until much later — more on that below.
Directors gave you their vision. You found the path.
A director would walk in with a picture in their head — the finished commercial, the hero shot, the feeling they wanted the audience to leave with. That was the entire brief, half the time. Not a process, not a pipeline, not a list of tools. Just: here's where we need to end up.
It sounds so simple, but there are a thousand ways to build that universe. Where does one even start? Working backwards from someone else's finished vision requires experience, subject knowledge, and a kind of deduction — running simultaneous permutations in your head, testing paths in reverse, until you find the one that actually gets you there.
I got a very literal version of this at Fin. We had a stadium to fill with a crowd for a commercial, and at the time, large-scale crowd simulation was still genuinely hard — the kind of thing only software like Massive could do (the same tool used for the battle scenes in Lord of the Rings). My lead, Stuart, looked at me and said, "You're learning Massive. The deadline is tight. Think you're up for it?"
Four weeks. Learn new software, fold it into our pipeline, deliver twenty-plus shots. The only way that's possible is by working backwards from the deadline and the end result, and figuring out the minimum viable path to get there — what to learn first, what to skip, what to fake, what actually matters for the shot versus what's just nice to know. You don't have time to learn Massive properly. You have time to learn exactly what this job needs from Massive.
It wasn't only about individual shots, either. Once Steve Oakley, another lead at Fin, showed me how to bend Maya to my will with MelScript, I stopped seeing our tools as fixed. I started seeing the outcome first, then working backwards through the software itself to find the shortest path to it — even if that path didn't exist yet. That's how I ended up building a frame buffer render management system at Fin, basically Maya's render layers before Maya had render layers. Nobody handed me a spec for that. I knew the outcome we needed and worked backwards to a tool that produced it.
It stopped being a technique and became a habit
Somewhere along the way this stopped being something I consciously did on a project and became how I approach pretty much everything. New problem, new industry, unfamiliar domain — doesn't matter. I picture where it needs to end up, and I work backwards from there.
Where it stops working
Here's the part I didn't understand for a long time: this only works when you already know what the end looks like. Reverse-engineering assumes the destination is fixed and fully formed, and the whole job is finding the mechanism that gets you there. It's powerful, but it's also bound by what you can already picture. You can't reverse-engineer a future you haven't imagined yet.
Qbism, the first game I ever shipped, didn't come from working backwards from anything. It came from staring out a car window at clouds on a drive back from the snow, daydreaming, no destination in mind at all. That idea moved forward, not backward — it emerged, and I chased it to find out what it was. If I'd tried to reverse-engineer my way to "make a game," I'm not sure I'd have landed anywhere near it.
So I've come to think there are two different tools here, not one. When the destination is clear — a director's vision, a deadline, a shot you can already see finished in your head — reverse-engineer it, ruthlessly. When it isn't clear yet, when you're staring at clouds, that's a different kind of thinking entirely, and trying to force backward logic onto it just kills the idea before it has a chance to become anything.
So which is it?
Do you actually know where you're trying to end up, in enough detail that you could work backwards from it? Or are you still staring at the clouds, waiting to find out what the destination even is?
If it's the first one: what do you need that you don't have yet? What's the minimum path, not the complete one? What would you learn in four weeks if someone told you the deadline was tight and asked if you were up for it?
What's your vision? Now you just need to work out how to get there.