This middle third is more willing to listen and to try to understand us, and it does not reject common-sense assumptions. One recent study6 found that people who held extreme positions on both sides of the climate change debate updated their views only with respect to information that supported their positions and not with respect to information that conflicted with their position. That is the bad news. The good news is that moderates in the same debate updated their views in light of information on both sides. They responded to evidence of all kinds. If this trend extends to other debates, then some arguments can reach this middle third by using premises that they accept, even if some extremists reject those premises. And reaching the middle third is usually enough to sway an election, if we are lucky, and then this moderate audience matters. In this way, arguments can often achieve important practical goals, even if these practical goals are limited, and even if there is no general theoretical reply to the challenge of the skeptical regress.
HOW CAN WE STOP?
We still need to figure out how to reach limited audiences with premises that they do not reject. In other words, we need regress stoppers for real life. Luckily, our language already supplies tools for this purpose. There are four main categories of regress stoppers: guarding, assuring, evaluating, and discounting terms. These groups of words can be seen as offering different ways to handle potential objections.
Guarding
Our first way to stop the regress is to weaken premises. To see how this works, imagine that you own a house in a low-lying area. A visiting insurance agent argues, “You should buy a flood insurance policy, because all houses in low-lying areas are destroyed by floods.” This argument is easy to refute, because its premise is false: It is not true that all houses in low-lying areas are destroyed by floods. Some survive. To guard against this objection, the insurance agent can weaken the premise to this: Some houses in low-lying areas are destroyed by floods. Now this guarded premise is true, but the argument runs into another problem: its premise is too weak to support its conclusion. If only one house in a million in a low-lying area is destroyed by a flood, then some are, but not enough to justify spending money on flood insurance. What the insurance agent needs is a middle path between a premise that is too strong to defend (“all”) and another premise that is too weak to support the conclusion (“some”). Here’s one intermediate possibility: Many houses in low-lying areas are destroyed by floods. This premise seems both true and strong enough to provide some reason to buy flood insurance. Of course, the term “many” is too vague to specify how strong this reason is (which affects how much you should spend on flood insurance). Nonetheless, the move from “all” to “many” improves the argument by avoiding some initial objections.
The same goal can be achieved by admitting uncertainty. Instead of claiming that your house definitely will be destroyed by a flood, the insurance agent could say this: “You should buy a flood insurance policy, because your house might be destroyed by a flood.” However, the fact that there is some possibility of a flood is hardly enough to justify buying flood insurance. If it were, then we would also have to buy meteor insurance, since any house might be destroyed by a meteor. A persistent insurance agent could try this premise in the middle: Your house has a significant chance of being destroyed by a flood. The vagueness in the term “significant” raises questions, but at least it makes the premise easier to defend and still strong enough to provide some reason for the conclusion.
These simple examples illustrate how guarding terms work. To change the premise from “all” to “many” (or “most”) or “some” or from “definitely” to “possibly” or “significant chance” (or “probably” or “likely”) is to guard the premise. Other ways to guard premises include self-description, as in “I believe (or think or suspect or fear) that your house will be destroyed by a flood,” since to object to this claim about my own mental state would be to deny that the speaker believes what he says he believes. How could we deny that? The purpose of all such guarding terms is to make premises less vulnerable to objections and thereby to turn bad arguments into better arguments and to stop the regress of reasons.
Despite their usefulness, guarding terms can be misused. One common trick is to introduce but then drop a guarding term. An insurance agent might argue, “A flood might destroy your house. That would be horrible. Just think of your cherished possessions. Your family could incur huge medical bills and would have to live elsewhere until you find a new home. In that case, our flood insurance policy will pay all of those expenses. Those costs add up to much more than the price of flood insurance. So flood insurance is a good deal.” What happened here? At the end, the insurance agent compares the costs of a flood destroying your house to the price of flood insurance. That comparison is relevant if your house will in fact be destroyed by a flood. However, the opening premise claimed only that a flood might destroy your house. If there is only a minute chance that a flood will destroy your house, then the costs of such destruction would need to be many times more than the price of flood insurance in order to make insurance worth the cost. The insurance agent has tried to hide this obvious point by dropping the guarding term. Watch out for this trick.
Another trick is to omit quantifiers entirely. People often say things like, “Houses in low areas are destroyed by floods.” Does this mean all, some, many, or most houses? If it means all houses, then it is false. If it means only some houses, then it is true, but not enough to support buying insurance. If it means many houses, then it is vague. Which is it? Until we know more precisely what this premise claims, we cannot determine whether the argument around it works. When someone tries to pull this trick, your best reply is usually: What do you mean: all, some, many, or most?
Let’s apply this lesson to a controversial political example. In early 2017, the United States stopped issuing visas to people from six Muslim-majority countries: Iran, Libya, Somalia, Sudan, Syria, and Yemen. Although the list of countries was modified later in 2017, let’s ask what kind of argument could support the original travel ban.
One common premise is simple: Muslims are terrorists. But what exactly does this mean? This premise is too vague to assess until we specify whether it refers to all, some, many, or most Muslims.
The first possibility suggests this argument: “All Muslims are terrorists. Everyone from these six countries is Muslim. Therefore, everyone from these countries is a terrorist.” This argument is obviously so bad that nobody ever presents it like this. Even the most ardent defender of the ban realizes that some people from these countries are not Muslims, and most Muslims (as well as most visa applicants) from these countries are not terrorists.
How can we fix this argument with a guarding term? One way is to weaken the premise from “All Muslims are terrorists” to “Some Muslims are terrorists.” This premise is easier to defend than the claim that all Muslims are terrorists. However, now it is too weak to support the conclusion. If we start the argument with “Some Muslims are terrorists,” then the premises will not be enough to support a ban on all people from these countries. How can we justify banning some political refugees who are not terrorists just because they happen to live in a country where some other people are terrorists? We need more justification for a ban on the whole country, so this premise has been guarded too much.
As with insurance, what we need is a middle path between a premise that is too strong to defend and one that is too weak to justify the conclusion. What about “Many Muslims are terrorists”? Is that premise strong enough to support a ban on everyone from these countries? I do not see how. One simple reason is that, even if many Muslims are terrorists, it still might be true that no terrorists come from these six countries. So at least we need a premise like “Many Muslims in each of these countries are terrorists.” Now, is that enough? Not yet, partly because the term “many” is so vague. Ten thousand terrorists is many terrorists. That means that, if ten million people live in a country, and ten thousand are terrorists, m
any people in that country are terrorists even though only one in a thousand is a terrorist. If we refuse visas to everyone from that country on the grounds that “many” are terrorists, then we ban 999 non-terrorists for every one real terrorist.
Maybe another kind of guarding term will work. It is true that every visa applicant from any of these countries might be a terrorist. However, it is also true that anyone from any country might be a terrorist. There is always some possibility, so a premise with the guarding term “might” cannot justify a ban on these countries without also justifying a ban on all other countries. Next, defenders of the travel ban could try this premise: “There is a significant chance (or too much chance) that any visa applicant from any of these countries is a terrorist.” However, some visa applicants have evidence that they are fleeing terrorism, so it is not clear why there is a significant chance that these particular applicants are terrorists. But then that guarded premise seems false.
Thus, it is hard to see how guarding in these ways could save this argument. Indeed, the fact that this argument is so dubious should make us wonder whether this argument is what proponents of a travel ban really have in mind. If we want to make fun of them, we might put such words into their mouths. But if we really want to understand them and their position, then we need to try to look at the issue from their perspective.
What other argument could they have in mind? One answer is suggested by asking why these six countries were singled out. It cannot be simply that they have Muslim majorities, since many other countries with Muslim majorities were not on the list. (Two non-Muslim majority countries—North Korea and Venezuela—were added to the ban later in 2017.) Instead, defenders of the ban claim that these countries’ governments are weak, corrupt, and chaotic, which makes it easy for terrorists to obtain false documents. Without trustworthy evidence, border officials cannot tell which visa applicants from these countries are terrorists. If even one in a thousand of those applicants are terrorists, and if we have no reliable way to tell which ones they are, then it is dangerous to issue visas to any of them. Whether it is too dangerous is another issue, but there is surely some danger in issuing visas without adequate evidence of safety. If this is the problem, then there is no need to guard the premise by moving from “all” to “some” or “many” or from “definitely” to “possibly.” The issue here is not the number of terrorists or the probability in a specific case but the unreliability of information about which visa applicants are terrorists. That lack of trust in available evidence explains why defenders of the ban want extreme vetting in all doubtful cases, and a complete ban when the political situation makes extreme vetting insecure or impossible.
I am not, of course, saying or suggesting that this argument is good or that it is bad. Evaluation is a separate task for later chapters, and it requires detailed factual information about the particular case. Here I am merely trying to determine which argument lies behind the travel ban so that I can understand why good and reasonable people support it and so that I can appreciate their reasons, learn from them, and figure out how to compromise with them. I suspect that at least some supporters of the ban have in mind something like this argument about trusting sources, but other ban supporters probably have in mind very different arguments. If so, then we need to figure out what their other arguments are and then try to learn from them and work with them.
Assuring
The issue of trust is addressed directly by a second way to head off questions and objections. Suppose that you wonder whether Sharif likes you, and I want to convince you that he does. I might say, “I assure you that he likes you a lot.” It would be impolite or at least uncomfortable for you to reply, “Your assurances are no good, because I do not trust you.” Thus, my assurance prevents you from objecting to what I say. But notice that I did not give any particular reason or evidence for my claim that Sharif likes you. I did not say that he told me that he likes you, that I overheard him praising you or saw him acting as if he likes you, or that a mutual friend reported such things about Sharif. When I say, “I assure you that he likes you,” I suggest that I have some reason for assuring you, but I do not openly specify what that reason is. As a result, you have no particular reason to object to. I also avoid saying how strong the reason is and how trustworthy the sources are. By specifying so little, my claim or premise becomes less objectionable and easier to defend. That is how assurance stops the argument and avoids a regress.
Instead of saying, “I assure you,” I could say, “I am sure” or “Surely,” “I am certain” or “Certainly,” “I have no doubt” or “Undoubtedly,” “There is no question” or “Unquestionably,” “Obviously,” “Definitely,” “Absolutely,” “As a matter of fact,” and so on. All such assuring terms suggest that there is a reason for a claim without specifying what that reason is. They thereby prevent the audience from asking for any more justification of the claim.
Assurances are perfectly fine in many cases. Some premises really are obvious, and sometimes opponents agree on certain premises as well as on the reliability of certain sources of information. It makes sense to say that evidence and experts support a claim without specifying any particular evidence or experts in situations where it would be pointless or distracting to go into more detail. Assurances can save time.
Despite these legitimate uses, assuring terms can also be misused. One common trick is abusive assuring. People often resort to excesses like these: “You would have to be blind not to see that . . . , ” “Everybody who knows anything knows that . . . , ” or (in the opposite direction) “Only a naïve fool would be deluded into imagining that . . . . ” Whenever people turn to abusive assurances like these, you should wonder why they adopted such desperate incivility instead of giving evidence for their claim.
Another trick is to allude to some source—authority or evidence—that you know your audience would reject without admitting that you are relying on this dubious source. Disputed reasons cannot resolve disputes. Imagine that a liberal watches a liberal news show (such as MSNBC) and says “Of course, the president colluded with our enemies” or “Anybody who keeps up with the news knows that.” These assuring terms do not explicitly mention the particular news source, so conservative opponents cannot object to this claim by criticizing its particular source. The same point applies to a conservative who watches conservative news (such as Fox) and says, “Only a dupe of fake news [or the mainstream media] would accuse the president of colluding with our enemies.” When assuring terms are used on both sides to refer to news sources that opponents reject, these assuring terms silence reasons on both sides because neither can discuss the reliability of unnamed sources. Such assuring terms stop argument, but they stop it too early.
Let’s apply these lessons to the United States travel ban discussed above. Imagine a visa applicant from Somalia or Yemen who says, “I assure you that I am not a terrorist.” An official whose job is to issue visas would have reason to doubt this assurance, since it is exactly what a terrorist would say. But then suppose an observer (perhaps another official or visa applicant) says, “She is unquestionably only trying to escape war and terrorism.” A visa official might trust this observer, but the rules might require reliable documentation. Even if the observer assures, “There is plenty of evidence that this visa applicant is safe,” the official would be well within his rights to ask to see that evidence. Then suppose the visa applicant produces what looks like an official document. Now the other side can resort to assurances. The official might reply, “That document is clearly unreliable. We know that documents like this are for sale on the streets of this country, and terrorists undoubtedly buy them.” These assurances give some reason for turning down the visa application, even though they do not say why the unreliability is clear or why the official knows about the sales and has no doubts that terrorists buy fake documents. That non-specificity leaves the visa applicant with no way to respond to the official’s skepticism.
The problem is that
assurances work only in a context of trust. If you tell me that you are certain, and if I trust you, then I might agree without needing to ask why you are certain. But if I do not trust you, then I will not be swayed by your assurance that you are confident or certain. Polarization often creates such lack of trust, so it undermines many attempts to share reasons, which breeds even more polarization.
Evaluating
A third way to stop arguments is to use evaluative or normative language. Philosophers have wrangled for centuries about the meanings of evaluative words like “good” and “bad” as well as normative words like “right” and “wrong.” I will not try to describe or contribute to those general debates here. I will only try to show how evaluative language helps to stop arguments in much the same way as assuring.
One venerable tradition suggests that to call something good is to say that it meets the relevant standards.7 An apple is good when it is crunchy and tasty. A car is good when it is roomy and efficient (as well as pretty, responsive, inexpensive, and so on). The standards for good apples are very different from the standards for good cars, but each is good when it meets the standards that are relevant to a thing of its own kind. Similarly, to call something bad is to say that it fails to meet the relevant standards. Bad apples are mushy or bland, whereas cramped gas guzzlers are bad cars.
Think Again: How to Reason and Argue Page 9