Science: Do It Yourself

by alyssavance 9y13th Feb 20115 min read207 comments


In the nerd community, we have lots of warm, fuzzy associations around 'science'. And, of course, science is indeed awesome. But, seeing how awesome science is, shouldn't we try to have more of it in our lives? When was the last time we did an experiment to test a theory?

Here, I will try to introduce a technique which I have found to be very useful. It is based on the classical scientific method, but I call it "DIY Science", to distinguish it from university science. The point of DIY Science is that science is not that hard to do, and can be used to answer practical questions as well as abstract ones. Particle physics looks hard to do, since you need expensive, massive accelerators and magnets and stuff. However, fortunately, some of the fields in which it is easiest to do science are some of the most practical and interesting. Anyone smart and rational can start doing science right now, from their home computer.

One of the key ingredients of DIY Science is to discard the more useless trappings of university science, for these frequently do more harm than good. Science doesn't need journals and universities. Science doesn't need beakers and test tubes. Science doesn't need p < 0.05, although I have found p-tests to be occasionally useful. The point of science is not to conform to these stereotypes of academia, but to discover something you didn't know before. (To our detriment, this is the opposite of how science is taught, as noted by Paul Graham: "So hackers start original, and get good, and scientists start good, and get original.")

Instead, as an simple first example, consider this question:

 - I want to get rich, or to be specific, have a net worth of over $100M USD. How do people get rich?

Here, we have an opportunity: We don't know something, and we want to find out what it is. To answer this question, our first intuition might be to Google "how do people get rich?". This isn't a horrible method, but by just asking someone else, we are not doing any science. Googling or asking a friend isn't the scientific method; it's the medieval method. (In medieval times, we would just have gone to the Church and asked, and the Church would have replied, "Pray diligently to the LORD and have faith, and you will be prosperous." Different people, same thing.)

In fields like physics, where lots of science is already being done by others, this will probably be OK. However, what if the question isn't about physics, like most questions people ask? Then, when you ask Google or a friend, you wind up with complete nonsense like this, which is the first Google result for "how do people get rich". Most people don't know how to use science, so that sort of nonsense is what most people believe about the world, which is why Western civilization is in such a mess right now.

Instead of Googling or asking someone else, we can apply the scientific method of actually looking at the data, and seeing what it says. Who are some rich people? How did they get rich? Where can we find information on rich people? The simplest technique, the one that I used when answering this question, is:

- Google the list of the Forbes 400.

- Go through each of the biographies for people on the list (or the first 200, or the first 100, or whatever is a large enough sample).

- Write down how they got rich.

- Summarize the data above: How do most rich people get rich?

Actually looking at data is simple, easy, and straightforward, and yet almost no one actually does it. Here's another one: Adjusted for inflation, what is the average, long-term appreciation of the stock market? Here's the historical Dow Jones index, and here's an inflation calculator. Try it and see!

The underlying principle here is very simple: Want to know whether something is true? Go look at the data and see. Look at the numbers. Look at the results. Look at a sample. JFDI.

For another simple example, one that I haven't done myself: It is a common perception that lottery players are stupid. But is it actually true? Is stupidity what causes people to play the lottery? It's easy enough to find out: look up a bunch of lottery winners, and see how smart they are. What jobs do they work in? What degrees do they have? What about compared with the average American population? What do they have in common?

There are an infinite number of these sorts of questions. How accurate are food expiration dates? How important is it to wear a helmet on a bike? How likely are STD infections? How many Americans are college graduates? Dropouts? What about high-income Americans?

Unlike most university science, DIY Science can actually make you happier, right here and now. One particularly useful group of questions, for instance, concerns things that people worry about. How likely are they, really? What are the expected consequences? What does the data say? For example, when I was younger, when I got a cold, I used to worry that it was actually some serious disease. Then, I looked up the numbers, and found out that virtually no one my age (10-25) got sick enough to have a high chance of dying. Most people worry too much - what things do you worry about that make you unhappy? What do the data say about them?

Or, suppose you want to save money to buy something expensive. The usual way people do this is, they take their income, subtract all of their necessary monthly expenses, and then figure that whatever is left over is how much they can save. Trouble is, people's necessities grow to match whatever their income is, even if their income is $2,000,000. If you get used to something, you start seeing it as "necessary", because you can't imagine life without it. How do you know if you really do need something? Use science! Try, just for a day, not using one thing with those monthly payments attached- electricity, phone, Internet, car, cable TV, satellite radio, what have you.

Of course, it isn't always easy, because sometimes people try to fool everyone. For instance, intelligence is distributed on a bell curve. Everyone knows that... right? As it turns out, the only reason IQ scores fit a bell curve, is because IQ is defined as a bell-curve-shaped statistic! Now, after the lie has been exposed, come the interesting questions: How is intelligence actually distributed? How could we find out? What measurements could we use?

Sometimes, questions get so politically loaded that you have to get tricky. To name a perennial favorite: Is global warming happening, and if it is, how much damage will it cause? It doesn't matter how much funding the NSF or some other agency gives this question, because the answers are already pre-determined; "yes" and "a lot" if you're a Blue, and "no" and "not much" if you're a Green. Peter Thiel, SIAI's largest donor, sums it up very nicely:

"There’s a degree to which it is just a status and political-correctness issue. The debates are for the most part not about the policies or about the ideas, but what is cool, what is trendy. Take something like the climate-change debate. I think it’s an important question, and I think it’s actually quite hard to figure out what the science is. It might be something for us to worry about. But I think there’s actually no debate at all — there’s no attempt to understand the science. It’s mostly moral posturing of one form or another.

Beyond the posturing, it’s a form of cowardice that’s very much linked to political correctness, where it’s not fashionable or not cool to offer dissenting opinions."

So, how do we really find out? Which evidence can we use? Where can we find it?

In exploring DIY Science, we ought to question everything, even things that we know (or think we know) to be true. "Common knowledge" is such a bad guide that false things float around for decades, all the time. Consider Wikipedia's List of Common Misconceptions. Reading through the whole thing, how many did you think were true? And these are the small set of things for which we have undeniable proof!

To name something which I do believe to be true: do men and women have the same average intelligence? They do, but how do we know that? Present studies can't be trusted, because the field is too politicized. You have to also look at pre-1970 studies, which indeed show agreement with modern ones. (Of course, past studies aren't always right, but agreement across many different time periods is fairly strong evidence.)

Or, to look at the subject of this blog: is rationality an effective means of achieving goals? To what extent? How do we know that? Well, on one side, what statistics I can find show that atheists make more than Christians. But they also show that Jews have higher incomes than atheists. Should we all convert to Judaism? Or, to take historical cases, Franklin was far more rational than average, but Hitler was far less. Clearly, more analysis is needed here.

One idea might be to look at what top chess players do: chess is a very objective metric, the players all have the same goals (to win the game), and the game is purely about mental decision-making. How rational are Garry Kasparov and Bobby Fisher? What about the top few hundred players worldwide? I don't have any clue what this will find, just a wild guess. 

I say all this on this blog, to some extent, because thinking about the data is not the only component of rationality; in order to have rational beliefs, one must also gather lots of data, and specifically, data about the problem one is trying to solve. No one in ancient Greece, no matter how well they thought, could have a good understanding of particle physics, because they didn't have any data on how particles behaved. Fortunately, with the Internet and online ordering of everything under the sun, data is very easy to collect. So- forward, in the name of Science!