1. Information is not to be trusted at face value, but instead quarantined and carefully scrutinized before acceptance.
Perhaps this seems a bit harsh or overdone. You might even say it seems...critical. This is where critical thinking takes its name: we are critical of all data, arguments, and conclusions. This criticism is the most important at the lowest level, which is data. All of the other levels (conclusions and the eventual decisions) are based upon data. If the data is bad, then even flawless reasoning will produce bad conclusions.
Our hypothetical space colony would not have much luck keeping out disease or spies by only spot-checking visitors here and there. Only by systematic investigation by a well-defined process which is the same every time can the safety of the colony be assured.
By the same token, a structurally-sound worldview is only possible by a careful and rigorous investigation of every piece of data that we intend to incorporate into that worldview. Consider yourself the overseer on the construction site for your own mind. You want to make sure that every support beam is whole and free of cracks or defects, that every concrete form is straight and aligned, and that every joint is secured properly with a sufficient number of screws. If any one piece of the structure is flawed, then integrity of the entire structure is suspect. The first earthquake that comes along may destroy the entire building.
This is a challenging task, without a doubt. Perhaps the hardest part is that we start "in the hole" already. If you are 25 years old when you begin utilizing critical thinking, that's 25 years of historical data which needs to be sorted through and verified. It can't be done overnight by any means. But the rewards of having a sound mental structure can do wonders for helping you achieve your goals in life. By basing your decisions on sound conclusions and sound data, you will make the right decision to get what you want far more often. Perhaps just as significant, you will rarely find yourself facing regret for having made a poor decision earlier on which shaped your later life - because you know you made the best decision that you could with the information that you had at the time.
The term "critical" tends to have a negative connotation for some people. Being critical of your own thoughts and ideas as well as those as others may feel like you're the wet blanket, always raining on people's parades. But think of it this way: a person who has fine taste in art is essentially someone who is critical about art. They carefully separate artwork that is poor, mediocre, and even pretty good, limiting themselves to only the very best pieces. They are respected for this: we say they have good taste. In the same way, a critical thinker is someone who has good taste in thoughts and information.
If you need further convincing of the importance of critical thinking, consider this. The nineteen men who hijacked and then flew two planes into the twin towers in September 2001 were most certainly in full control of their mental facilities. They had the ability to forge documents, pass themselves off as passengers, pilot the planes, and a host of other challenge tasks. Yet they each had come the grossly egregious conclusion that the death of thousands of people as well as their own deaths would benefit them in some way. They believed perhaps that it would assure them a place in heaven, or that it would improve circumstances for their family and countrymen. These conclusions may have even made sense in the context of the data they possessed. But the data they had was wrong. If these men had thought critically about the information they had and the conclusions they were basing their decisions on, they would never have proceeded with such a horrible crime.
Obviously most of us will never reach such extreme conclusions or make any decisions like this. But the principle is the same for the more common cases of deciding whether to get married, what career path to take, and which home to buy. The decisions you make today will have long-lasting effects on your life; why find out fifteen years from now that you choose the wrong career or the wrong house? Critical thinking, properly applied, means that you will usually make the best decision given the information you have available to you at the time.
So, how do we change our thinking about incoming data? A number of points must be considered every time new data is encountered:
Is it really data? Secondhand information is almost always an argument, or in many cases a conclusion without any supporting data. This type of information is borderline useless, with its only potential value just the awareness of that person's position. For example, if someone tells you that you should vote for Smith for the office of mayor in the upcoming election, you would have to let them know that you need some supporting data. What they have offered you is a suggested decision based on an argument which does not contain any supporting data. An argument without data is a conclusion, and conclusions can only be generated internally by you, not inserted into your mind by someone else. Therefore you would reject this information outright, as it contains no data. (Well, it does implicitly state that Smith is running for mayor, and that the person that told you is supporting them. But this data is not very relevant to supporting their argument, which is that you should vote for them.) In a world of critical thinkers, bumper stickers and signs commanding you to vote for a particular candidate would be nigh-on worthless.
How does it fit in with the data I already have? Assembling a worldview is like working on a giant jigsaw puzzle. The pieces of objective reality fit together in only one way. However, our data is provided by our very limited and imperfect perceptions. The totality of all reality is too vast for a single human mind to hope to contain, nor is it really necessary or desirable to do so. Therefore our worldview is like a puzzle into which we are constantly fitting pieces. Each new piece of data that comes in is like a puzzle piece, a chance to further expand our picture just a little further. But more often than not, we are handed a piece which goes to a different puzzle, or that goes to no puzzle at all. So when we try to add it to our puzzle, we find that it won't fit anywhere.
This one comes naturally to most people. For example, if someone tells me that today is Wednesday, I may reject this data fairly quickly, because 1. I remember that yesterday was Monday; 2. I looked at my calendar this morning and it showed Tuesday; 3. I spoke to a co-worker earlier who mentioned that it was Tuesday; and 4. I looked at my watch just now and saw that it was Tuesday. Therefore I reject this new piece of data, that today is Wednesday, as false. We say that the preponderance of evidence supports the statement that today is Tuesday.
Of course it's usually a lot harder that this. For example, if the mechanic presents you with the argument that you that you need to change the timing belt in your car immediately because it's about to break, you will try to fit that in with other data that you already have. You might remember changing the belt only a few months ago, and that the belt only needs to be changed every 40,000 miles. Therefore his new data (that it needs to be changed) does not fit with your existing data (it was changed recently and only needs to be replaced infrequently). You will most likely challenge him with this data. He will then try to show additional data to support his case, such as opening the car's hood and showing you that there is a tear in the belt. He may then also offer the general data that sometimes belts are damaged by rocks getting caught under them. You will then probably try to fit this in with data of your own: for example, have you driven on a gravel or unpaved road recently? Ultimately you will try to avoid coming to a conclusion (whether or not the belt needs to be changed) until all your data fits together reasonably well.
Keep in mind that it's not just the incoming data that can be thrown out. Your existing data can, too. A simple example: your friend tells you that Los Angeles has extremely cold winters. But when you move there yourself and experience your first winter, you determine that it's not that cold compared to other places you have been. Therefore, you throw out your previously held data (LA has cold winters) in place of two new pieces of data (LA has mild winters, and your friend is a wus).
In other words, data is not on a first-come, first-served basis. Your friend gets the front seat when he calls "Shotgun!". Data, on the other hand, is evaluted on merit, not the order in which it arrived.
Often data cannot be verified by direct observation. Returning to the belt example, you might have thought that your belt was replaced a few months ago, but in fact the mechanic you took it to at that time scammed you: you were charged for the belt, but he did not replace it. You didn't actually observe him replacing the belt, so you can never verify this for sure. But you might combine the data that you did not observe the belt being replaced with other data such as "I remember that guy seemed kind of shady to me" and the current mechanic telling you that the belt looks extremely old and worn, far older than a few months in his opinion. Based on this data and the speculative conclusion that you were mistaken about the belt being replaced previously, you may throw out your old data (belt was replaced a few months ago) with new data (belt was not replaced at that time, I got ripped off by paying for a service I did not receive, that mechanic is untrustworthy). This new data is of use not only in deciding your current course of action (approving a replacement of the belt) but also in making future decisions (do not return to the shady mechanic, perhaps warn friends away from them).
What is the source? Although this is not a reflection on the pure data itself, it does provide a simple shortcut for deciding how much immediate weight to give the data. For example, if a friend of yours who was a skilled mechanic took a look under your hood and told you that your belt needed to be replaced, you'd probably assign a lot more immediate weight to that data than you would if another friend who had no particular knowledge about car repair told you the same thing. We also tend to lend extra trust to sources that have always guided us well in the past, and extra suspicion to sources who have purposely or accidentally misled us in the past. There is also the matter of partiality: a mechanic at a repair shop is skilled, but they stand to benefit based on which decision you should choose to make (you are deciding whether or not to buy their services). Therefore we tend to balance our trust of their expertise with suspicion of their motives when considering the data they provide.
Although useful, source knowledge is only a quick-and-dirty guide that you can use to quickly determine whether a piece of data is worth further investigation. People are often misled by trusting someone simply because they are experienced - a mechanic is human and can make mistakes and misjudgements like anyone else. Another common fallacy is to assign blanket trust to a person who has shown their knowledge in one field, and then misapply that trust to another. For example, the minister of your church may give excellent spiritual guidance, but his advice about what house to buy or whether to get your timing belt replaced is unlikely to be better than anyone else's.
How can I validate the data? First, is it validateable at all? "The sky is blue" is verifiable by going outside on a clear day and looking up. "Your timing belt has a tear" is verifiable by looking at the belt. Most data is harder to check, like "there were 16 robberies in our city last month" or "alcohol damages your liver." The first example is the type of data that is only available to use secondhand; no one in the world can possibly have first-hand knowledge of this, because they would have to be in every place in the city simultaneously for a month. Instead, we get this data by compiling reports of robberies submitted to the local police department. So to verify it, you need to check with the source that compiled the information: perhaps the police department has posted this information to their website, or you can call and speak to someone there that can confirm that is their official number, and how it was arrived at.
In some cases, simple consistency tests will allow you to throw out data. Consider a newspaper article which claims a new study shows that 200 million Americans are at risk for breast cancer. You may find that this data fails a consistency check against other data (breast cancer only affects females, there are 150 million female Americans) and can therefore discard it without even getting into the details of the study.
"Alcohol damages your liver" is one of the hardest types of data validation problems that a person is likely to face. Since direct inspection of one's liver is quite difficult, and since the amount of damage done by a single serving of alcohol is so tiny as to be imperceptible, one has to turn to the more abstract secondhand knowledge produced by scientific studies of the subject. The studies are long-term and have a great number of variables involved. Many factors must be considered when looking at the results of such studies, including how they were conducted, whether the sample set was large enough, and how variables were normalized in the study (for example: were all the people studied also smokers, in which case it might be the smoking which caused the liver damage?). There is also the simple matter of how the data fits in with your personal situation. You may find, looking at the studies, that liver damage is only a problem for people who consume three or more glasses of alcohol per day, every day, for at least 30 years. If you are a person who drinks only on occasion in a social situation, you can then split this data into two relevant parts: "alcohol damages the liver when consumed excessively over an extremely long period of time" and "alcohol's damage to the liver is imperceptible when used only infrequently and in moderation."
Many times we will find ourselves in a position where a piece of data need to be momentarily accepted in order to consider all of our options, or for other reasons we want to accept it temporarily. For example, listening to someone giving a speech which presents a broad thesis including includes numerous of arguments and supporting data. Temporary acceptance is a part of life because you cannot always verify everything on the spot, but it's important that we keep it marked as tentative data inside our minds. Like the security guard escorting a visitor through the space colony, we have reason to accept it for now, but it's important that we not be lulled into thinking that this temporary acceptance is permenant. In fact, manipulative people may use this as a trick to get you to accept bad data. They present some data and ask you to assume it is true for the moment so that they may move on to present their conclusion. Their conclusion seems compelling and you accept that it makes sense given that the premise is true. But you haven't verified the premise, you only accepted it temporarily for the sake of their argument! Therefore, you must mark both the data and the conclusion as tentative until you can verify the data.
Last of all, realize that data acceptance is always colored by its level of verification. Tentative data is the most obvious example, we must keep in mind the method by which and how thoroughly it was verified for all data that we remember. For sometime in the future we may encounter a piece of data which doesn't fit into our puzzle, and then we have to decide: which data will we keep, and which will we discard? In order to determine this, we must compare correctness, which means comparing their methods and levels of validation. For example: your doctor advises you that drinking alcohol in any amount will cause serious liver damage and impair your health later in life. You accept this data and the conclusion that goes with it, because your doctor is a knowledgeable and trustworthy source. But perhaps later you visit France, and see an entire culture whose members drink a small bit of alcohol with each meal throughout their lives, and has many healthy seniors who live to ripe old ages without any liver problems. Furthermore, you do some research online and discover a number of studies from credible sources showing that alcohol is only harmful to the liver in consistent, high doses. So now you have two conflicting piece of data, one historical and one current, and both have achieved a minimum level of verification. So which to use? Although both were verified, the first piece of data came only from a single source. Although that source was trusted, you know that individual people, even doctors, can often make mistakes or push agendas of their own using their trusted status. On the other hand, your observation of the widespread use of alcohol in French culture compared to the absence of late-life health problems, and your research into the studies of the long-term effects of alcohol, provides a broader and stronger level of support to that piece of data. Therefore you may conclude that social drinking is not a problem, and decide to have that glass of wine at the company Christmas party.