Before the printing press, there were very few books. It was extremely hard to learn math; you had to pay a tutor. Of course literacy was very low — but all knowledge that could be transmitted through books (such as math) was very low.
Science cannot be taught through books. You can learn a lot about calculus by reading books. You can learn almost nothing important about science. Science is not a collection of facts, it is a method, a way of gathering knowledge. Almost always it is taught by doing — by working in a lab, for example. Just as, before printed books, almost no one could do any math, it is true today that almost no one can do any science. (Most doctors think the bigger the sample size, the better.)
If you look at a biology textbook, it is full of conclusions. It says practically nothing about the process by which those conclusions were reached. For some reason biologists have decided not to teach that — perhaps because it is difficult and messy to teach. And someone might be offended. Whatever the reason, the process goes undescribed. And it’s all sciences, not just biology. (Until recently, economists avoided teaching data. At least in introductory economics, data was too messy for them.)
As long as you have to learn science by doing it practically no one will understand it — just as almost no one did math when you had to hire a tutor to learn it. But now we have the Internet. And blogs. Two new things have entered the picture: a great deal of emotion (blogs are full of emotion, unlike textbooks); and unlimited space. Now science can begin to be taught without actually doing an apprenticeship. If you add enough emotion, anything becomes riveting. And there is now plenty of room for all the false starts and messy details. I suppose most scientists who blog are too worried about being dignified to say anything emotional or messy, but that doesn’t matter because there are so many bloggers.
According to Stephen Dubner, “if you are fan of science, this [Climategate] is a pretty grim day.” I think it’s a great day. As great as the day the first math text was printed. It’s the first time a large number of people are getting a real lesson in science. Mainstream media coverage is pathetic but there are so many bloggers it doesn’t matter. You can read about it endlessly. As you do, you will painlessly and unforgettably learn what Leonard Syme taught his students for years, and what I blogged about a few weeks ago: The apparent consensus on any difficult issue is more fragile than it looks. You are learning how conclusions are actually arrived at. It isn’t pretty — which textbook writers and professors, seeking dignity above all else, fail to mention.
Speaking of the scientific process, I thought you’d appreciate this: “Brotto did a bit of experimenting on herself. Not that she suffered from any disorder, but she talks of herself sometimes in a researcher’s terms as ‘an n of one,’ a single subject on whom she likes to test her ideas.”
Seth wrote: “If you look at a biology textbook, it is full of conclusions. It says practically nothing about the process by which those conclusions were reached. For some reason biologists have decided not to teach that — perhaps because it is difficult and messy to teach.”
Or perhaps because they did not follow the right method to get to those conclusions.
According to Dennis Prager, the purpose of graduate school in the social sciences is to destroy the student’s ability to think for himself. Ouch.
Hi Seth,
As a doctoral student expecting to learn science (albeit in an engineering department), I was dissappointed. My advisor (though I think he knows how to do science, wasnt interested in making things explicit, perhaps he doesnt care about me?).
I think I learnt a lot about Science by reading blogs (especially radical and controversial blogs) which I never learnt by reading books. Maybe because most books (with rare exceptions) are interested in just the final result, while blogging is an interacting medium where a blogger might be questioned and he has to respond to comments.
BTW you are also amongst those who helped me in ‘learning’ Science. So just a thank you!
@Seth
Fantastic post. The hidden and unappreciated gem in your post is as follows:
“Science is not a collection of facts, it is a method, a way of gathering knowledge.”
99.9999% of people (high priests, er, I mean, “scientists”, included) do not understand this differentiation.
Science is a method, not a body of knowledge.
When I hear people say “the science tells us X, Y and Z” — I ask myself, how does the method “tell” us anything? You could have one universe of data and facts interpreted an infinite number of ways, all potentially valid.
The reason most people conflate science as a method with science as a body of knowledge is that the former is harder to control (being that the principle tenet of science as a methodology is skepticism), while the latter becomes a religion and a political tool of the elites.
The reason most people conflate science as a method with science as a body of knowledge is that the former is harder to control
I would rather say the reason is that it is harder to define!
May be you shoud read a little philosophy of science, no consensus here either.
Very interesting ideas.
But as well as input, you need a feedback loop. The apprenticeship system provides for constant monitoring of the apprentice to check that the information really is being understood. Otherwise you could be endlessly misunderstanding.
Andrew, I’m not sure what you are referring to when you say important stuff about science can be learned from reading the Feynman Lectures. I read Volume 1. I like it, but it taught me nothing about how to do experiments or analyze data, which is 99% of the science I do. I can’t imagine it affects your work. So what are you talking about?
milieu, any blogs you would recommend to learn about science from? yeah, the fact that blogs deal in radical and controversial stuff makes them much easier to learn from.
Patrik, thanks. I think people who do interesting research know perfectly well what I say is true, I just think it would make their lives more difficult to acknowledge it — how crappy the usual textbooks are, for example. So they don’t. I wouldn’t say the primary attitude of science is skepticism. I think it is to think for yourself, which is quite different. To think for yourself can mean being more skeptical than everyone else; it can also mean being less skeptical than everyone else. To think for yourself means reaching conclusions based on whatever evidence is available, as opposed to reaching conclusions based on what other people think. Perhaps that is the core reason science (the method) is rarely taught — because it would give whoever is teaching you less power over you.
“To think for yourself can mean being more skeptical than everyone else; it can also mean being less skeptical than everyone else.”
Excellent!
thanks, Bruce. Medical Hypotheses embodies that idea much better than most other journals.
Well, I think my ‘eyes opened’ when I read blogs like gnxp gnxp , Eric Raymond’s Blog , Overcoming Bias . I do not agree with many of their ideas but I learnt that the established truths are not sacrosanct and have to be questioned.
But I would admit that the starting point of all these might be the book “The Black Swan” by Nassim Taleb and his writings on the web including his website .
These are not per se Scientific but another one might be Eric Drexler’s blog , again with a heavy technology bias.
Andrew, thanks for explaining that. I’ve learned a little bit about data analysis from reading — from reading Exploratory Data Analysis by Tukey, in particular. But since that book didn’t include examples of real-life data analysis, what could be learned only went so far. Sort of like what you could learn about cooking from a cookbook that used only two ingredients. To say that you learn science when you learn how to do a t-test (or any statistical method) is true, but I think it confuses many people. They confuse the detail with the big picture. The big picture is that doing science consists of deciding what data to collect (what experiment or survey to do), dealing with difficulties along the way, extracting the most info out of the data you’ve collected, and seeing how what you’ve learned fits with what people already knew. For some people it also consists of raising money and effectively working with students and/or employees. From Tukey I learned a little bit about value of transformations and graphics. Deciding what data to collect is by far the most important step and, in my experience, is never taught by books.
milieu, yeah, I think Taleb does the best job among living writers of teaching that you shouldn’t take established truths as so established. And he describes himself as a philosopher, not a scientist!
Is “On the Origin of Species” worth reading? Which edition? The 6th, as the final, most finished version? Or the 1st, as the edition that had the most historical impact?
It has been a long time since I read it, but my recollection is “Chaos: Making a New Science” by James Gleick might qualify as a book that one can read to learn a bit about how science is done.
One story I remember in particular was about Edward Lorenz studying atmospheric models and unexpectedly finding widely divergent results from simulations that started with nearly the same conditions (as I recall, it had to do with roundoff while re-starting simulations). His investigations into this led to the idea of the “butterfly effect” or a bit more formally, “sensitive dependence on initial conditions.”
Matt, a friend of mine said Chaos by Gleick is a good example of journalistic misunderstanding. Gleick thought the science was much more important than my friend thought it was. As far as I can tell my friend was right — that stuff has gone nowhere. The big decision for most scientists is what data to collect. What you describe is remote from data collection.
I don’t feel like I have enough of a comprehensive understanding of the field to know about the big-picture importance of what Gleick specifically wrote about in Chaos, but I certainly would not argue with the idea that “Chaos Theory” and fractals had their 15 minutes of fame in popular culture where they were “the next big thing” that was going to “revolutionize our understanding” and then they faded from public view. That said, it is also my understanding that the study of ‘chaos theory’ (by which I mean non-linear dynamics and complexity, going back to the late 1800s) has resulted in some important breakthroughs (though perhaps you would not consider them to be scientific).
As I understand it, one significant idea to come out of these studies has been that for many even relatively simple systems it is not possible, even in theory, to directly calculate the system state at an arbitrary point in time. Even worse, it is often not even possible to accurately simulate future states of a given system.
Another idea that I think came out of this field is the understanding that a given system (even one that is completely deterministic) may have some parts of its parameter space where it is ‘well behaved’ and others where it is ‘chaotic’. This idea suggests investigations to understand where/how the transition between “nice” behavior and chaotic behavior occurs.
My background is mostly mathematics with a healthy dose of statistics. I do not consider myself a scientist, so perhaps my understanding of what is included in science is faulty. However, I tend to think of Lorenz’ work as being scientific in nature. I am curious about how you would define/describe science. For example, what role, if any, does mathematical modeling play in science?
Finally, for what it’s worth, BBC did a program called
“High Anxieties – The Mathematics of Chaos” (available on youtube) which appears (I’ve not watched the whole thing yet) to get into some of the impacts of this field of study. Of course this doesn’t necessarily mean that studies are scientific in nature.
Matt, my guess is that 99% of published scientific articles involve data collection or summarizing (review articles). That is their emphasis. They may fit lines or other functions to the data, sure, but that has nothing to do with the stuff in Chaos. Mathematical modeling of other people’s data (no original data collection) is among the 1%. Articles about mathematical modeling of hypothetical data are very rare.
“Mathematical modeling of other people’s data (no original data collection) is among the 1%.”
Isn’t that what all the climate science is about, creating models based on data? Sure, most of the articles may be data collection, but the heart of the argument for AGW is based on models.
Maybe the stuff in chaos has been a dead end because the lesson is “most systems are too complex to be modeled so as to generate useful predictions.”
Lance, data, not models, are at the center of Climategate. For example, failure to provide data after it’s been asked for. The models can’t be taken seriously because they haven’t predicted anything correctly, as Bruce Charlton says. I think the heart of the argument for AGW was stuff like the hockey-stick graph. With that gone, I suppose people will shift to saying it is the models that matter.
So you’re saying the billions spent on climate science are primarily devoted to collecting data and not to run models on expensive computers?
Lance, no I am not saying that. What’s your point?
I’ll answer with a story (you love stories, right?) In ninth grade, I did a science project on chaos models, the simple equation f(x)=x(1-x), which can model the swings in rabbit populations, for instance. I spoke with one of the judges who told me something I haven’t forgotten: non-linear systems are very hard to disturb — after some internal shock they will almost always revert to the original equilibrium.
You slighted mathematical modeling, but the models are crucial for the case. Data is one thing, but the models tell a great story. And of the 21 models followed by the IPCC, all 21 assume that increasing C02 will lead to changes that amplify warming.
https://www.drroyspencer.com/2009/07/how-do-climate-models-work/
But if one paid attention to non-linear systems and models, then you would be very sceptical of a) the ability to predict temperature based on models and b) models that assume a system as the earth’s climate is unstable. That may be a negative conclusion, but negative conclusions are still conclusions.
“Non-linear systems are hard to perturb.” For non-linear systems with negative feedback that’s true; for non-linear systems with positive feedback that’s false. The climate modelers say the system has positive feedback (“amplify the warming”).
>> is true today that almost no one can do any science. (Most doctors think the bigger the sample size, the better.
There are a number of things that are very hard to learn without personal teaching (usually involving building things) but this example isn’t onw of them./
Every book or article about polling will tell you this. I’ve read this many many times, and it is very logical. Once you get beyond about 1,000 increasing the size of the sample doesn’t add much to the accuracy – the key thing is the quality of the sampling. If doctors don’t know taht that;’s only because they never read anything about it. perhaps they had to take calculus, but not statistics – actually a statistics course might obscuire this tye of thing – just read any old book about political polling.
>> If you look at a biology textbook, it is full of conclusions. It says practically nothing about the process by which those conclusions
Textbooks are probably about the worst place from which to learn anything.
Now to know something about how conclusions were reached you need to read Asimov and other wroters. \This is usually found at around Dewey Decimal number 500 or 504 or 508 in the library. You need collections f essays or any popularization.
One problem: Books like “Microbe Hunters” can be misleading because since 1948 or so we’ve had the problem of “peer review” and “double blind studies” and grants. This is misleading as to how science works nowadays.
A good science fiction story dealing with how science could be controlled is “The Dead Past” by Isaac Asimov. This aspect of the story might not be captured in the summary:
https://en.wikipedia.org/wiki/The_Dead_Past
Asimov’s belief wa sthat in the long run (which could mean decades) science couldn’t be suppressed because one field impacts on another
>> The apparent consensus on any difficult issue is more fragile than it looks. You are learning how conclusions are actually arrived at.
This is stuff everybody should know. And nobody will do good work without knowing it. And neitehr should policy be made by people who don’t question the work.
It’s harder for people to realize because so much is developed now so it looks better than it is.
Talking about chaos, if anybody thouyght a little bit about they would realize that IF it were true that there is a tipping point where a sluight increase in CO2 could start a runaway greenhouse effect (which is actually what the alarmists are touting) it would most likely be achaotic system so that whether it happened or not would depend n just where exactly the value was – it would be something wiotha starnger attarctor, so that ti would make very little sense to lower the level of CO2 slightly – you ciould actually start the runaway greenhouse effect that way – you;d be dealing with the probability of it happening, and it wouldnt be worth it to go to a lot of effor to reduce the probability slightly.