Simulated Annealing And Using Randomness To Fuel Creativity
Simulated annealing is an algorithmic process used to solve complex mathematical problems. The ideas used in simulated annealing can be surprisingly helpful in fuelling creativity when you feel stuck or are unsure what to do next.
Creative ideas sometimes seem to come from nowhere. We describe this as being inspired. But inspiration is a fickle source of motivation. Chuck Close famously said, “Inspiration is for amateurs; the rest of us just show up and get to work.”
The problem isn’t believing in inspiration. The problem is waiting for it. Waiting for inspiration before you try to create anything is a recipe for procrastination. If you work only when inspired then you’ll get very little done, take forever to finish anything, or give up thanks to the frustratingly slow process.
So we push past how we feel and just get on with the task of making and creating. Often, as Charles Baudelaire suggested, inspiration flows from doing the work.
But sometimes we get stuck. Maybe we feel like we’ve run out of ideas. Perhaps the problem we’ve taken on seems too difficult. Or maybe the only paths we can see ahead require more skill than we currently have. What do we do next?
The answer may come from a technique used in maths and programming – simulated annealing.
What Is Simulated Annealing?
Annealing is the process of heating and then cooling a metal to change its properties and make it more workable. It can make metal easier to shape or bend, or stabilise it after it has been worked or welded.
Simulated annealing takes this idea of heating and cooling and applies it to complex mathematical problems. It’s particularly useful for the kind of problems where you have a large number of possible answers and must find one that is optimally efficient.
Let’s say you have 20 parcels to deliver all across town today. You want to devise an efficient route. But you can’t spend all day working it out, and you don’t necessarily need the absolute best route. After all, traffic conditions might change during the day. You just want to be sure you’ve rejected all the worst routes and chosen one of the better ones.
Now imagine a bigger version of that problem. Maybe you have 20 trucks in a big city, and hundreds of parcels. Or 20 cities, each with 20 trucks. Or 20 cities in each of 20 countries. That’s the kind of problem simulated annealing can help to solve.
There’s three interesting features to these kinds of problems. They are complex. They must be solved promptly. And finding the perfect answer doesn’t matter – you need only to find a good enough answer and to avoid the bad ones.
Why This Matters For Creative Work
Consider the kind of problems we face with creative work. There’s a lot of options. It’s hard to choose between them. But some are clearly worse than others and should be avoided. The obstacles we face are sometimes unknown. The clock is ticking and we need to make something.
And we could procrastinate forever, trying to find the perfect answer.
Imagine you’re writing a book, or making a film, or recording an album. At the beginning you don’t know exactly the shape of your final product. There’s no single correct version you can imagine before you start. Its final form will be the result of the many decisions you will make along the way.
Choosing a path still matters. Some choices could be catastrophic. Others might invite delays or frustrations. Some could ensure you never finish.
In order to begin, you don’t need the best answer. You need an answer that’s good enough to get you going. This is important to remember whenever you feel stuck or overwhelmed.
Try A Little Randomness
So, how does simulated annealing work? Remember the heating and cooling process in metalwork? In simulated annealing the heat refers to throwing a lot of guesses at the problem without concern for their accuracy. The cooling starts when some of the guesses start to reveal a pattern.
The heating is fast and violent. The cooling is slower, taking longer to calmly identify the best guesses, especially as you start to see the better solutions.
As Brian Christian and Tom Griffiths put it in Algorithms to Live By,
“…Simulated Annealing: you should front-load randomness, rapidly cooling out of a totally random state, using ever less and less randomness as time goes on, lingering longest as you approach freezing.”
Start Anywhere – Focus Later
The cliche says “You have to start somewhere.” The lesson from simulated annealing is that when you’re stuck, you can start anywhere. The important thing is to start (we explored something similar in the article Before Planning Begin) Start, and the best alternatives will reveal themselves.
So, the next time you are stuck, try anything. Literally anything. Try lots of anything. Move quickly between different types of anything.
Then, as you start to get ideas that feel like they work, slow down. Be less random and focus more. Pick an idea that feels good enough and commit to it. Trust that your decision has led you out of the place where you were, and will lead you forward.
And keep going.