Being wrong about stuff is both fun and easy. There’s a unicorn in my garden who brings me ice cream every day! See, you can’t tell me that’s not an improvement in every way over the sad reality of my actual life.
However, some people aren’t happy with this idea. Some people don’t want me to have a unicorn. Some people are more interested in being able to distinguish true things from untrue things, and only want to believe the former. Some people want to take their ideas about how the world works, and then improve them over time, as they learn more stuff. They say that this leads to a “better understanding” of the world, and has provided us with useful things like “technological advances” and “improved quality of life”. Whatever good that‘s supposed to be.
It’s difficult to know where to start to explain why the scientific method is a good thing, because it seems like it ought to be enough to wave my hands around and go, “Well… duh!” It really does seem that obvious that this is a good way of doing things, and actually articulating an argument in its favour seems almost unnatural. And yet, not everyone sees it as a self-evidently good thing, so explaining its usefulness is important.
So, sarcasm off for a moment, as I try to describe more or less how science works.
Firstly, people notice things that are going on. Everyone does this, even if they’re not doing science. We wouldn’t be active participants in the world if we weren’t always observing things, processing them, and deciding how to act based on our interpretations. For instance, it has been noticed for centuries in most parts of the world that the sun appears at one horizon, moves across the sky, and sinks below the other horizon, at a rate of once per day.
After noticing a few things about the world, we might come up with some interesting questions as to how it works. These questions might look like: “Hey, you know how the Sun rises and sets every day? What’s up with that?”
Once we’ve found a question to ask about the world, we can start coming up with answers. At this point, pretty much anything that answers the question, and explains whatever phenomena we’re asking it about, is a potentially good next step, and is called a hypothesis. It might be solidly based on previous research, or it might be some crazy shit we came up with while we were stoned and staring at our hands with a profound sense of wonder. For now, it doesn’t matter.
Noticing something, asking a question about it, and proposing a hypothesis, might look something like this:
How does that great fiery ball move across the sky each day, providing us with light and heat? Perhaps the great god Helios drags it behind him in his chariot.
My friends ate the berries from that bush, and then soon afterwards they made choking noises, fell over, and stopped moving. Why did this happen? Maybe they were God’s berries, and he struck them down for stealing them.
My friends ate the berries from that bush, and then soon afterwards they made choking noises, fell over, and stopped moving. Why did this happen? Maybe there was something bad in the berries that’s harmful to be eaten.
Why does everyone point and laugh at my mullet whenever I go outside? I guess nobody round here has any fashion sense.
We humans are immensely complicated creatures, and we live in a fantastically complex and beautiful world. How could all this wonder have come about? It must have all been put here by God.
And so on. It’s often not verbalised quite so formally, but this process of thinking is the basis of formulating hypotheses.
Next we start to come to the real meat of the scientific method. Using our hypothesis, we start to make predictions. We say: okay, if this idea we’ve suggested is really how things are, then it explains what we’ve already noticed, but what else should follow? What else should we see, if we keep looking at things, and maybe dig a little deeper? And, perhaps even more importantly, what doesn’t follow from our hypothesis? What do we not expect to see?
This last bit is vital, and demonstrates a crucial way in which science differs from non-scientific and pseudo-scientific approaches to the world. We basically gave ourselves free range to be creative with our hypotheses, which is great – creativity is important in science – but it can lead to some pretty wacky ideas. If our friends died after eating some berries, then angry gods and poisonous fruit both provide a line of cause and effect that explains it just fine. But if we don’t go any further, there’s no reason to think that any one hypothesis is “better” than any of the numerous others we could have picked. We have to see whether we’ve picked a good one, by doing some hypothesis testing.
If an explanation is going to be any good to us, it has to be specific enough to predict what we’ll see when we look in certain places. And hand-in-hand with predictive power comes falsifiability – if our hypothesis predicts that something will happen, then there must be some other things which, according to the hypothesis, shouldn’t happen. If they do, then our hypothesis is a bad one which fails to fit the evidence.
For instance, our hypothesis about the berries might simply be, “These berries are poisonous”. This explains why the people who ate them are now dead. One prediction it makes about the future is that anyone else who eats the berries should also die shortly afterwards. We could put together an experiment by which to test this hypothesis, such as feeding the berries to someone we don’t like and watching to see whether they keel over. (Cruel, perhaps, but it’s FOR SCIENCE!) If they did, this would support our hypothesis.
But if they didn’t, then our hypothesis has a problem, and may need to be abandoned. However proud with ourselves we may have felt for coming up with this brilliant explanation, it might be bunk. If it fails in its predictive powers then we can’t afford to keep clinging to it just for old time’s sake.
The idea of falsifiability may seem odd, or not really that important. If your theory is good, then why should you need to be able to prove it wrong, in order to prove it right? The thing is, unless there’s some imaginable way that it could seem wrong, it doesn’t really tell us anything interesting about the universe.
There could be an invisible, intangible, inaudible, and very mischievous imp living in my wardrobe, which would perfectly explain what keeps happening to my socks. But if this imp is completely undetectable, then this tells me nothing about what I’m likely to observe in the future, and he may as well not be there at all. If, on the other hand, I know something specific about this particular breed of imp, then I can make predictions like “If I leave these socks out here, they should disappear at a certain rate”, and I can potentially find out if there’s no invisible imp after all, if I keep good track of my socks and they stay put.
Then, once we’ve noticed some new things, and gathered some new data (whether in a lab experiment, or just by looking somewhere different, or whatever), we check how well the hypothesis holds up.
If things happened like we predicted they would, yay! Looks like our hypothesis has some usefulness. We’ve successfully predicted something with it. It might even be a good description of how the universe is. That’d be neat. Once this has happened a few times, and we’ve started building up a substantial and well-established model of what’s going on, we might start to call this hypothesis a theory.
If they didn’t, then maybe the hypothesis needs tweaking a little bit. Maybe the imp only likes green socks, or the berries only poison people during a full moon. Depending on the exact nature of the results, we might come up with a slightly different, better hypothesis, which explains these new results as well as the old ones, and which does predict things correctly the next time we gather more data. But it might just be that it was a bad hypothesis, and we should give it up and think of something new. In the above cases, it’s probably more likely that there is no invisible sock-stealing imp; and maybe my dead friends ate something other than the berries as well, as it seems unlikely that the lunar cycle would have such an effect. (More on Occam’s Razor in a future essay.)
And, crucially, it’s a never-ending process. Once you have a theory, which can explain things and usefully predict the future, you keep testing it, you constantly watch out for new evidence, or perform new experiments, to see if it holds up, to make sure you really are as right as you can be, and to leap on any possible shortcomings or failings in your current model. And if you find some, then you come up with something new and go through it all again.
This is why science rocks. If you’re doing it right, you will always, always be learning new things. Your understanding of the world will get better and better, because you’ll be putting all your ideas out there for people to test, and they will be trying their damnedest to pick away at any flaws and tear your models down, to prove you wrong, over and over again – and when they find they can’t do that any more, and it seems that you absolutely must be right, whatever facts they gather and whatever experiments they run, then you know you’ve got as close to the truth as you can possibly get. And then you still keep looking.
It’s win-win. If you were right all along, then nobody will be able to use any facts to prove you wrong, and the more they look into it, the more it’ll look like you’d got it sussed from the start. But if you were wrong, either completely or in some small detail, then when it starts to look that way – when enough evidence turns up which your hypothesis can’t explain, and when it’s not predicting the future as accurately as some other model – then you get to change your mind and be right anyway.
Science rocks. The scientific method is the best set of tools we have for minimising our collective wrongness. Use it. Be righter.