Stop optimizing your life

-May 10, 2016

We all know the articles and books titled along the lines of ‘How to be more effective at planning’ or ‘at work’ in general, ‘at making friends’, at ‘making money’. Then there are the works explaining how to increase happiness in your life or decrease stress.

In short, we have a tendency to optimize our life. To increase the amount of meaningfulness, to minimize the amount of sorrow and to make every minute count. We want to be able to design our life.

This article talks about how difficult optimization problems are and what a terrible idea it is to apply it organizing a human life. I think science, to be very general here, will figure out a way of optimizing lives. But we are talking about right now, and not by science but by altering your own life.

When I did my PhD in theoretical nano-optics I did may fair share of trying to solve optimisation problems. That is why I know it is very hard, even for reasonably well-defined mathematical models let alone a complex organism such as ourselves living in a complex society. By discussing several concepts I will try to explain to you why it is hard and why we should do without.

Global and local minima

You can understand the act of optimizing your life as walking around in a hilly landscape. What you want is to reach sea level. You could try to find the highest mountain peak but you do not know how high that may be, so it is safer to always aim for the 0, sea level. Now changes in your life map to walking into a certain direction. Say you want to lose weight and cut down on your fat intake. By eating less and less fat you are walking down the hill of being overweight, if you are lucky. If you ate more fat, you would be walking uphill again, so you don’t. At some point cutting fat doesn’t help anymore and you have reached the valley.

But you are nowhere near sea level and you’re not satisfied yet. The problem is that from where you are now you first have to climb up a hill again to see if the valley behind that hill is closer to sea level, and there is no guarantee that lower valleys lead to the sea level. We are stuck at a local minimum. It is the best we can obtain in this area. It is not the global minimum, the absolute best solution there is.

It would be nice to just have a map of the whole world so we can just jump to the right place. The problem is that so many parameters determine the landscape that it is very hard to map and I am pretty sure that for each person the map would look different enough so that a general map would be useless. Given that people’s physique and daily life is so different, it is no surprise that diets, changing only a part of your life, will work for some and will not work as well for others.

In summary, making slight changes in your life will make things a bit better or a bit worse but probably keeps you longing for that magic solution, the dream diet, but that will be nearly impossible to find for us regular folks.

The algorithm

Mathematically you can find many ways to look for that magic solution, the global minimum. The techniques are called algorithms and I will not bore you with all the options. Generally, there are two main types. First are the descent type of algorithms. They work pretty much like I described above. You are on a hill, you look around you to see what direction goes down the steepest and you start walking. If you take very small steps you may be afraid to step over a molehill and stop there, if you take seven-miles leaps you may jump over that coveted valley. Mathematically you can analyse your surroundings even closer to determine the best direction you walk in, but in general you end up in a nearby local minimum. To find the global minimum you just have to start yourself in many different random places on the map and just try out how it goes. In practice, you won’t have the time nor the ability to just place yourself somewhere randomly on the map which would mean you would have to completely change your lifestyle in some random way.

The other type are evolutionary or genetic algorithms. This works similarly to nature. In our analogy of hiking in the hills this would translate to the following story. We are now following 1000 people who are standing around in the landscape. We pick the 100 people that are closest to sea level, because they did the best. We let them make babies. By mixing their superior lifestyles we hope that their babies (we introduce some random mutations) will have similar lifestyles and will eventually produce a baby with the super dream life style. In a way we take more random leaps in the landscape.

Again, for an individual trying to optimise his life this is quite useless. Science might study the entire population and the lifestyles they have but in the end it would be difficult to identify what choices you would have to make in your life to reach that dream state.

The merit function

I did not even discuss the most important aspect of optimisation, the merit function. What are you even optimizing? If you follow the advice to be a more effective leader, what does that do to your health? How does becoming more healthy relate to your efficiency in planning? You would first need to define exactly what you think is important to optimise before you start to follow random advice online. You might think you can separate losing weight, being happy and being effective at work. Sadly, we have conservation of misery.

Conservation of misery

During my PhD project I had to rewrite my math in all kinds of forms to make the equations easier to solve or to be optimised. You can do that actually, you can make the equations more elegant to solve one thing but it will be a complete mess to solve the deeper thing you actually also needed to know. My professor coined this to be the universal law of conservation of misery. If something is complex or nasty, it will show itself however you try to make it look more elegant from one perspective.

The alternative

So it seems hopeless. What can we do? It mostly happens automatically. We are learning machines by nature. If we see a handy tip and it is appropriate for us we will try it out and make our lives marginally better, or sometimes worse and we throw that tip out of the window. Remember everyone is near a local minimum and all that tips can do is bring us closed to that local minimum.

You can choose to accept this minimum but sometimes you just can’t. If you are in such a bad place you need to get out, literally. Take the escapist light solution. Most people do it already, take a vacation. Go to a different culture, follow a different daily routine and eat differently, but also, think differently, speak differently. Do not try to bring your old life into the vacation life. Try out this new place on the map. Some people escape their old life for good, leaving wife and children behind for a new life. Those people should remember that the life they enter is also a local minimum and will not be perfect. It would be better to return to the old ways and make adjustments based on your vacation experiences. In genetic algorithm terms, create a baby you from the old and vacation you and see how that goes. It won’t be ideal but at least it should take you away from your current hillside. It might mean breaking with certain friends, quitting your job or moving, but it might be worth it. Remember the global minimum is impossible to find, deal with it.

Humans and optimisation

Ok, so what I have shown is that it is highly unpractical and unrealistic to optimise our lives as we would like. But I want to take it one step further and call it unwanted as well. Optimisation works for clearly defined processes. These processes cannot think for themselves and have no feelings. The moment humans (or animals to some degree) enter the process we have to be careful. By prescribing what humans have to do, we take away their autonomy to make their own decisions, to try out new things and learn from them, to act based on feelings. As imaginative idealists we have to take those considerations into account and that means it would be wisest not to optimise processes involving humans. Instead, we should set clear but spacious boundaries and allow for freedom within those boundaries.