After I wrote a little essay about life being better without planning, one of those fortunate coincidences occurred, and I found a book called “Why Greatness Cannot Be Planned”.
I recommend the book. It may not convince you, but it presents a perspective that even sceptics could usefully consider.
It makes the case, from a technical and practical base, that no ground-breaking achievement can ever be achieved by deciding on the steps to get there and working through them. That it’s not the way nature works, and not how progress happens.
If somebody visited from the far future, and told us that time travel is possible and has been invented by then, that would be useful information, possibly. But how would we plan to invent a Time Machine as a result? With current knowledge it’s too radical a leap. Like looking across a foggy landscape to a distant peak rising out of it – there it is, but the route is completely unknown.
Solving problems by not knowing what you’re doing
The central idea of the book is organised around problem-solving algorithms, the main illustration being finding a way through a maze.
You might assume that this is reasonably straightforward, and there will be a set of rules a robot could follow that would ensure it will find the exit from any arbitrary maze it’s presented with. But this turns out not to be the case, and once the complexity rises to a certain level, even the best algorithms have quite low success rates. This in spite of knowing that success is possible.
What does work, with far greater success, is what the authors call novelty-seeking.
New for the sake of it
Leaving technicalities aside, if you program a robot to always try to do something it’s never done before – and that’s all, without giving it any goal at all – it turns out that it will solve more problems.
In a maze, for example, hitting a wall is repetitive. If a robot’s only aim is to do new things, hitting a wall won’t get it far. Nor will blind alleys. Anything that results in a dead end will be tried and rejected for future tries … but the crucial thing is that the robot has no reason not to try anything in the first place. So it isn’t misled by promising false starts, and it doesn’t avoid routes that look as though they go in the wrong direction but turn out to be a long way round to where it needs to be.
If a route leads to increased possibility, it will be explored and favoured, because there is more newness that way. The ultimate being the exit of the maze.
The same applies to robots learning to walk. Give a robot legs and actuators, and the control process for stable walking is difficult to design. But train it using novelty-seeking, and it will quickly exhaust the possibilities of staying on the floor, or even falling down, and will find ways of moving further and further in as reliable ways as are possible.
Obviously nobody has programmed nature, but the authors make a good case for treating evolution as a sort of novelty-seeking model.
The downside
There is solid evidence, then, for this approach being extremely powerful.
The problem? If your only explicit aim is “something new” and goals are banned, then outside narrow carefully designed experiments, there’s no telling what you’ll get. As the authors say, if you ran evolution again on Earth from the beginning, the chances we’d end up with humans again is vanishingly small. Given the way evolution favours ever-increasing complexity (the main reason they argue that it is like a novelty-seeking principle) we can be fairly sure that something remarkable would turn up, but nothing as specific as humanity.
So if you must have a particular outcome, you’re out of luck.
Mind you, as the authors point out, if you want a particular outcome and the path to it isn’t clear then you’re probably out of luck anyway. Better to try the novelty route and get something even if it wasn’t what you hoped for.
OK, but is this leading anywhere?
This isn’t meant to be a book review as such. I just need to cover this ground to make the case for what otherwise looks like aimless novelty worship.
Outside of robotics, it has general application to anywhere in which we want to see “progress”. In our personal lives, our careers, in business, the economy, society generally. And it works best if we can accept a vague definition of “progress” – that we want things to be better than they are now, but are not set on a vision of what we think “better” looks like.
(Which, incidentally, fits with another post of mine about learning that what we think will make us happy usually doesn’t – living cheaply by choice).
The practical application
What this tells us is simple, if hard to accept for some:
If there is something within reach that we think is worth achieving, then plans will get us there. For anything else, we’re more likely to end up with something worth having by not even trying.
That’s not to say we do nothing, only that we don’t try to achieve something in particular. Novelty-seeking has its own rules, and they are quite explicit.
Don’t do exactly what has failed before. (Obvious, but difficult for most of us.)
Judge your new thing not by what you get out of it, but what it opens up. This is subtle but important. Doing something new won’t achieve something spectacular in itself. The only criteria is whether or not it opens up fresh new things that weren’t possible before. If your action revealed more possibilities, that was a good one. If there is nothing more, it wasn’t.
For humans, “interestingness” is the most valuable guide. Humans are curious. Why? Who knows, but evolution has made it part of our make-up, and what we sense as interesting seems to be a proxy for promising novelty. Don’t rationalise it away – if something feels intensely interesting, that’s your instinct that it leads to new paths, and new paths are what counts.
Don’t just take my word for it, either. Think of the most fulfilled people you know, and consider how much they live their lives this way, knowingly or not.
Consider how many scientific and technological breakthroughs have come this way, from “pure research” that nobody knew would lead anywhere. Arguably there were a greater number of radical advances in times when there was R&D funded without expectations or targets.
The caveat, of course, being that we do need to have security to be able to do this, either personally or as an organisation. Few of us can do whatever we like in the hopes something will pay off – the bills need paying first. But it should be the first thing we strive for once we can.
The pernicious precautionary principle
This brings me to a modern idea that does incredible damage to all of us by now – the precautionary principle.
The precautionary principle says that if something is untested, it should be assumed to be dangerous until proven otherwise. And it is the legal basis for a lot of regulation, and the first thing applied to most innovation.
Which sounds extremely sensible. Common sense. “Better safe than sorry”.
But it’s insidious, because it’s completely incompatible with the novelty-seeking approach that is provably the only way things can get radically better than they are. The route to unknown future benefit relies on an ever-increasing branching of paths, most of which we can’t tell the benefit of at the start. If we trim the number to only the ones we can prove are safe, we slash our chances.
Treating any new approach as a risk is the absolute opposite of what we should normally do.
We can see this all around us.
Places in the world where new things are treated as risks tend to fall behind, economically, places where new things are embraced.
Businesses that stick to what they know are sooner or later overtaken by competitors with new ideas.
People who settle into routine have mid-life crises.
We protect our young people these days, and the jury is out on whether or not they’re benefiting from that.
Going back to economics and competition, it explains a familiar pattern. A country or business may be ahead of others, able to do things they can’t and therefore profit enormously. Others see what they’re doing, copy and catch up, then even pull ahead by doing it even better, and the original country or business panics. But progress isn’t linear, and this is as far as the planning mentality can take the challengers … to a point they could see was possible, a step within the bounds of the known. Usually, the next radical change that leads to new dominance comes from somewhere entirely new and unexpected, from somewhere that was doing something nobody thought was important at the time.
Governments and company boards never seem to learn this, though.
So what am I saying here?
Firstly, I am convinced we are too confident that we know what’s good. We should ease up and be more open-minded to see if there might be happy outcomes from things we don’t expect.
Secondly, to whatever extent we practically can, we should do new things. Vary things. Try things out. Do whatever hasn’t been done before, ideally by anybody, but at least by us.
Thirdly, pay attention to whatever is interesting, and don’t let a lack of obvious use put you off. Or other people’s lack of understanding.
Finally, be very wary of anyone or any system or organisation that puts risk assessment ahead of anything else, unless the stakes justify that approach (and don’t be sucked into thinking that justification is easy, either). Safety assumes predictability, and nothing great ever came of anything predictable. Never support campaigns that aim to eliminate risk unless you’re also comfortable with eliminating progress in that area.
Mostly, though – do new things just for the sake of it. You’ll be happier. And more successful, probably, in ways you couldn’t have predicted.