The internet is rife with advice to conduct further research into claims before accepting them. The phrase ‘Do your own research’ is a widespread and infamous piece of advice, especially popular amongst conspiracy theorists, crypto investment advisors, and alternative health influencers. Even people who wouldn’t say “Do your own research” will say things in the neighborhood, offering similar advice under a more respectable guise. For example, some who wish to combat the spread of false information advise people to employ lateral research techniques, searching the internet to see whether or not they can find other sources that bolster the credibility of the initial source or confirm its content. This, too, is advice to do your own research.
The phrase ‘Do your own research’ (often abbreviated “DYOR”) has been widely criticized. In particular, philosophers have criticized the use of ‘DYOR’ to instruct one’s audience to employ their own critical reasoning and engage in inquiry rather than deferring to experts. When those who question the views of mainstream experts (hereafter, “fringe theorists”) advise their audiences to do their own research, the advice especially suggests that traditional sources of information are not trustworthy. ‘DYOR’ erodes trust in the scientific establishment, mainstream media, and other relatively reliable sources of information. In other words, ‘DYOR’ advises you not to defer to experts the way you ought (Levy Reference Levy2022). I agree this is a bad feature of many uses of the phrase ‘DYOR.’
However, in a particularly interesting set of cases in which speakers tell their audience to do their own research, the speakers also present their own view to their audience. Such cases call our attention to one of the striking features of ‘DYOR.’ When someone tells their audience to do their own research, they don’t just direct listeners not to defer to the experts; they direct listeners not to defer to the speaker themself either.Footnote 1 Take, for example, a speech by Mark Steele, warning listeners that the primary purpose of 5G and LED street lights is to serve as a weapons system. After explaining the purported dangers, Mark Steele instructs his audience to do their own research, elaborating, “Don’t take any notice of me. I do not want one person in this room to believe anything I say. Not one word of it.”Footnote 2
The use of ‘DYOR’ to instruct people not to defer to the speaker about the matter at hand may be subject to less scrutiny because it seems like good advice. After all, ‘DYOR’ is especially popular among people espousing beliefs that run counter to the mainstream view.Footnote 3 In most cases, the audiences of such individuals should not regard them as reliable and should not defer to them. Proponents of lateral research and fringe theorists agree you should not take the fringe theorist’s word for it; you should do more research. Therefore, conspiracy theorists, hacks, and grifters telling their audiences not to defer to them may seem like a purely positive thing.
However, it is important to see that instructing someone not to defer to you may have rhetorical effects other than just encouraging your listener not to defer. There is something apparently self-defeating about instructing your audience not to defer in the context of making assertions about your view. Usually, people make assertions hoping their audience will take their word for it or at least be persuaded by their arguments. Appending ‘DYOR’ apparently instructs listeners to do their own research rather than defer to the speaker. Even stranger, it instructs listeners not to form a belief on the basis of the evidence the speaker has presented – rather it instructs listeners to leave the question open for further inquiry. Why bother presenting your view while telling your audience not to accept your view on the basis of your presentation? The purpose of this paper is to propose two possible answers to this question and, in doing so, to bring out further reasons to be suspicious of advice to DYOR. First, I want to address and set aside one legitimate reason why people might bother making an assertion while telling their audience not to defer to them. Some bad actors make confident assertions, fully intending their audience to take their word for it, and then append ‘DYOR’ in order to escape accountability for the reliability of their assertion (Ballantyne et al. Reference Ballantyne, Celniker and Dunning2024). Such speakers expect their audience to defer to them, but when their audience comes around to blame them for asserting something irresponsibly, they want to be able to say, “I told you explicitly to do your own research rather than trusting me.” For ‘DYOR’ to serve this function, it is not important that the audience pay any mind to the instruction to do their own research at the time of utterance. The speaker only says “DYOR” as a form of verbal “fine print” they can refer back to when they need to avoid accountability. This is a common and criticizable use of ‘DYOR.’ However, I am interested in understanding what could be gained by using ‘DYOR’ if you expect your audience to pay attention to your instruction when you say it and take you to mean it sincerely.
I will argue that telling one’s audience to do their own research can be a way for people with low credibility to bring their audience around to accept their view. The phrase ‘DYOR’ is thus, in this usage, somewhat paradoxical; on its face it discourages people from believing the view presented, but its effect is to encourage just that. After telling his audience not to believe a word he says, Mark Steele continues, “It is important you do your own research because then you can absorb it better. You can realize what I’m talking about and you go, he was right.”Footnote 4 My goal is offering a possible explanation of why telling audiences to do their own research makes one’s view easier for the audience to absorb. I will also discuss the implications of these uses of ‘DYOR’ for more mainstream versions of the advice to do your own research, like the advice to employ lateral research.
Before moving forward, I will make a few notes about the scope of my project. Throughout this paper, I will be making claims about the rhetorical purposes that advising your audience to do their own research could serve. These claims will naturally raise the question of how often advice to DYOR actually serves these rhetorical purposes. How often do people who give the advice to DYOR intend for their advice to have this effect? How often are people who hear the advice to do their own research actually influenced in the way I describe? These are both empirical questions, and my ability to answer them here will be somewhat limited. I aim to make it plausible that the advice to DYOR is sometimes intended to have the effects I describe and does sometimes succeed at having those effects. Where I am aware of empirical evidence to support these claims, I will provide that evidence. However, I take the arguments I make below to have interest even if people are aiming to have these effects relatively rarely and even if the number of people affected is relatively small. Let me say more about why.
First, someone need not intend to achieve a rhetorical effect in order to actually have an effect. It is quite plausible that many people who use the phrase ‘DYOR’ are simply mimicking others in their community and have no further objective. That does not mean the phrase won’t have further impacts on the audience. It is, in fact, fairly common for strategically minded subgroups within fringe communities to introduce some rhetorical device with the intent that the broader community will unwittingly adopt and spread the device (Marwick and Lewis Reference Marwick and Lewis2017). Even if the vast majority of people who advise their audiences to do their own research have purely non-rhetorical intentions, the possible rhetorical functions of the phrase are still important.
Second, these strategies need not be effective for a large proportion of the audience in order to be of interest. I think it is fairly plausible that, relative to the total audience, few people are impacted by the advice to DYOR in the way I describe. After all, most of the audience will already be convinced that the speaker’s theory is right or that it is wrong (or completely uninterested in the question). However, I am particularly interested in why people who come to accept fringe theories wind up in that situation. I am not suggesting all or even most listeners are impacted in the ways I describe below. I am making the more modest suggestion that the phenomena I will go on to describe are part of an explanation of how people can come to accept fringe theories. Even if most people who hear a fringe theorist tell them to do their own research dismiss the advice out of hand, I am still interested in how the advice affects those who take it seriously.
One final note. Some of the literature about ‘DYOR’ understands doing your own research to require fairly intense research procedures with high degrees of intellectual self-reliance like gathering or processing raw data. There is good research to suggest that within some fringe communities, the standards for doing your own research are, in fact, that high (Lee et al. Reference Lee, Yang, Inchoco, Jones and Satyanarayan2021). These practices invite the common criticism of the advice to DYOR that most non-experts are not equipped or inclined to perform this kind of research reliably (Ballantyne et al. Reference Ballantyne, Celniker and Dunning2024). If your options are to defer to the experts, engage in this kind of ambitious research, or suspend judgement, engaging in this kind of ambitious research is downright epistemically irresponsible.Footnote 5 I agree with this criticism. However, since I am focused on how advice to DYOR is understood by people who have not already bought into a fringe theory, I will not be focused on the kind of intense research practices that can be found within fringe groups. When most people hear advice to do their own research, they do not immediately track down raw data. They search the Internet.Footnote 6
As mentioned, I will argue that advising their audience to do their own research is a way for people with low credibility to bring their listeners around to believe their view (or at least, raise their credence in the view).Footnote 7 ‘DYOR’ helps those with a credibility deficit bring their listeners to accept their fringe views in (at least) two ways, by signaling their epistemic virtue and by creating an illusion that their audience has found more independent evidence for their theory. First, I will explain how the advice to DYOR functions as a virtue signal to increase the speaker’s credibility. Second, I will explain how fringe theorists direct their audience to do their own research in order to give their audience the illusion that they have found evidence independently. This illusion gives the evidence more apparent probative force. In actual fact, the fringe theorist has directed their audience’s research towards confirmatory evidence.
The first section addresses ‘DYOR’ as virtue signaling. The second section defines what it means for research to be independent in the colloquial sense. The third section explains the evidential value of independent research. The fourth section describes how fringe theorists steer their audience’s “independent” research while maintaining the illusion of independence. The fifth section uses case studies to demonstrate this phenomenon in action. The sixth section concludes.
1. Virtue signaling
There has been a flurry of recent literature on the phenomenon of moral virtue signaling.Footnote 8 Moral virtue signaling is an action a person takes in order to make others believe the agent is a morally virtuous person.Footnote 9 So, for example, if someone wears a ‘Stop Hate’ pin in order to make others believe they are compassionate, they are signaling their moral virtue. If someone volunteers at the soup kitchen in order to make others believe they are beneficent, they are signaling their moral virtue. While virtue signaling is usually defined as an intentional action, we also perform actions that unintentionally function as a virtue signal. Some virtue signals may also be virtuous actions. For example, the soup kitchen volunteer may manifest beneficence while also intentionally signaling their beneficence to others.Footnote 10 Certainly, not all virtue signals are also virtuous actions. Wearing the ‘Stop Hate’ pin is probably not itself an act of compassion. Furthermore, it is possible to signal that you have a virtue that, in fact, you lack. Arguably, billionaires who widely publicize their charitable giving are signaling that they are generous, but if their donations are paltry relative to their accumulated wealth, they are not, in fact, generous. Finally, an action may succeed as a virtue signal while failing to be a virtuous action if the audience has false beliefs about what’s virtuous (Levy Reference Levy2023). For example, in a community that believes servility to be a womanly virtue, a woman may signal “virtue” to her community by an act of self-abnegation. The virtue of her act is suspect because servility’s status as a virtue is suspect.Footnote 11 Even if servility is not a virtue, she may successfully lead her audience to believe she is virtuous by acting servile because her audience believes servility to be a virtue.
Recently, several authors have discussed how people signal their virtue in the epistemic domain (Levy Reference Levy2023; Harris Reference Harris2025, and Saint-Croix Reference Saint-Croix2025). Epistemic virtue signaling is what we do when we take an action in order to make others believe we are an epistemically virtuous person. If our pin-wearer puts on their ‘Trust Science’ pin to make others think they defer appropriately to experts, they are signaling their epistemic virtue. A scientist who appears on a cable TV network and uses domain-specific jargon in order to demonstrate their expertise is signaling their epistemic virtue. Most relevantly for our purposes, when people advise their audience to do their own research, they may be signaling their epistemic virtue.
What “virtue” might ‘DYOR’ signal? For starters, it may signal that the speaker is intellectually independent (Levy Reference Levy2023 and Harris Reference Harris2025). In advising their audience to do their own research, the speaker expresses commitment to the importance of independent research. If they are telling their audience that independent research is the way to proceed, the audience might well assume that the speaker uses that method of inquiry, too. While epistemologists have recently stressed our epistemic inter dependence, questioning the possibility and desirability of doing research that is purely your own, they level these critiques against a background societal presumption that doing your own research is valuable.Footnote 12 Even if independence is not a genuine virtue, a speaker can virtue signal by signaling independence so long as their audience believes intellectual independence is a virtue.
‘DYOR’ may also signal that the speaker is especially diligent and has mastery over the relevant subject.Footnote 13 Why might ‘DYOR’ signal diligence and mastery? Consider the use of ‘DYOR’ in an adversarial context. Rather than explaining their reasoning, a speaker tells any skeptics to go see for themselves. The implication is that the speaker is certain the available evidence will convince the audience of the speaker’s view because it overwhelmingly proves the speaker right. Onlookers may expect speakers would only tell skeptics to do their own research if the speaker has done the due diligence and developed enough mastery of the subject to be confident that their audience will not return from their research with even more reasons why the speaker is wrong.
Another virtue that may be signaled by telling others not to defer to you is epistemic humility. How is it possible that the same phrase can signal mastery or humility? It depends on context. For example, sometimes people will conclude their presentation with something like, “I’m not an expert. Do your own research.”Footnote 14 By drawing attention to their lack of expertise, the speaker acknowledges their limitations. This practice bears special interest because it directly undercuts the usual criticism that ‘DYOR’ undermines trust in experts. In these cases, speakers are not necessarily undermining trust in experts – they are primarily (apparently) undermining trust in themselves. However, their very instruction to their audience not to trust them may, in fact, signal that they are trustworthy. Humility is an epistemic virtue that counteracts the very widespread epistemic vice of overestimating the scope of your knowledge and the reliability of your reasoning. Someone who is epistemically humble is aware of their possible errors and shortcomings when forming beliefs and owns those limitations (Whitcomb et al. Reference Whitcomb, Battaly, Baehr and Howard-Snyder2017).
If advising your audience to do their own research succeeds at bringing an audience member to believe that you have one of these epistemic virtues, your credibility for that audience member will increase. Perhaps the ‘DYOR’ virtue signal will raise your credibility enough that the audience member ends up accepting your view after all. Even if the signal does not make a big enough difference for the listener to accept your view now, it may contribute to a longer project of increasing your credibility with your audience. Recall that our question is, why go to the effort of making assertions while telling people not to take your word for it? We now see one reason is that telling people not to take your word for it may raise your credibility such that they actually will take your word for it now or in the future.
However, someone might wonder whether telling people to do their own research is actually efficacious as a virtue signal.Footnote 15 Do audiences actually come to believe that speakers who tell their audience to do their own research have these virtues?Footnote 16
It’s important to distinguish the question of whether some particular behavior is an effective virtue signal from the question of whether it is a reliable virtue signal. In order to be an effective virtue signal, the behavior must lead the audience to believe (or be more confident) that the speaker has the relevant virtue. If some behavior is not an effective virtue signal, anyone who engages in the behavior in order to virtue signal is going to fail to achieve their aim. Insofar as people realize that some behavior that they are engaging in with the aim of virtue signaling is not effective in doing so, we should expect them to stop engaging in the behavior in order to virtue signal. Conversely, in order to be a reliable virtue signal, the behavior must actually be good evidence that the speaker has the relevant virtue. A signal could be effective but not reliable if the audience accepts the signal as evidence of virtue even though it is not good evidence.
Some may doubt that ‘DYOR’ is a reliable virtue signal and, on that basis, doubt that ‘DYOR’ is an effective virtue signal. There is a robust literature in evolutionary theory about what type of signaling displays are effective that connects effectiveness to reliability. According to evolutionary theory, we can expect an effective signal to be designed to be credible – that is, some element of the signal itself is a reliable indicator of the signaled property (Henrich Reference Henrich2009). One might object that we should expect effective virtue signals to involve a credibility-enhancing display, but ‘DYOR’ does not involve a credibility-enhancing display; therefore, we should not expect ‘DYOR’ to be an effective virtue signal.
I have two responses to this objection. First, in some cases, I suggest ‘DYOR’ does involve a credibility-enhancing display. One way that a virtue signal may involve a credibility-enhancing display is if the signal is self-ratifying.Footnote 17 A virtue signal is self-ratifying if, in order to make the signal, you have to employ the relevant virtue in the first place.Footnote 18 For example, volunteering at the soup kitchen may necessarily involve some beneficence on behalf of the volunteer. If so, the fact that someone volunteers at the soup kitchen directly demonstrates their beneficence. It might require some humility on behalf of the speaker to say, “I am not an expert, so do your own research.” If so, this version of the ‘DYOR’ virtue signal does involve a credibility-enhancing display. However, I agree that, in other cases, ‘DYOR’ is not a particularly reliable signal of epistemic virtue. Therefore, we must turn to my second response.
My second response is that the argument from evolutionary theory tells us what we should expect when the signalers and receivers are successfully pursuing their interests, but they often fail to do this for extended periods of time. The principle from evolutionary theory suggests that reliable signals are more likely to get uptake and persist (that is, be effective) because everyone benefits. The signaler is perceived as virtuous, and the receiver gets good evidence about the character of the signaler. Signals that do not involve a credibility-enhancing display are less likely to be effective because if the receiver accepts the signal, signalers can signal dishonestly and take advantage of the receiver.Footnote 19 The signaler is perceived as virtuous, but the receiver may be misled about the character of the signaler. It is not in our best interest to treat non-reliable virtue signals as evidence of virtue. However, that does not preclude us from doing so. We may fail to recognize that a signal is not reliable, or we may trust that it is not being used dishonestly. While reliable signals may be more likely to be effective, we cannot simply infer from the fact that a signal is not reliable that it is not effective.
This evolutionary theoretical point is not the only reason why someone may question whether telling people to do their own research is an effective virtue signal. Given the common critiques of ‘DYOR,’ one might wonder whether anyone still views the advice to DYOR in a positive light. Certainly, there are many articles that call into question the epistemic virtue of those who tell their audiences to do their own research. However, there are also plenty of people with large platforms who advocate doing your own research. Some use the phrase ‘DYOR.’Footnote 20 Some express a similar sentiment in other terms.Footnote 21 Either way, these figures reinforce the idea that a high degree of intellectual independence is an important virtue. While hopefully the critiques of ‘DYOR’ will eventually make the advice ineffective as a virtue signal, I think it is overly optimistic to believe that has happened already.
One might wonder whether mainstream speakers could and should benefit from telling their audiences to do their own research as a virtue signal.Footnote 22 I would suggest mainstream speakers ought not attempt to adopt the ‘DYOR’ virtue signal because doing so might reinforce the idea that a high degree of intellectual independence is an important virtue.Footnote 23 By engaging in a virtue signaling practice, you express your commitment to the relevant “virtue.” If we care about the character of the epistemic community, we should refrain from endorsing (and thereby reinforcing) false virtues.Footnote 24 The costs of reinforcing bad practices and values probably outweigh whatever benefits come from improving your own credibility. To the extent that ‘DYOR’ endorses an unvirtuous degree of intellectual independence, no one should use ‘DYOR’ to signal intellectual independence. While mastery and humility are genuine virtues, there are other ways to signal that one has mastery or humility that do not risk also reinforcing the idea that a high degree of intellectual independence is a worthwhile ideal.
2. Following the advice to DYOR
So far, I have been explaining how telling your audience to defer may actually increase their trust in you, making them more likely to defer to you (rather than doing their own research). Going forward, I will be exploring the possible rhetorical benefits for speakers if their audience actually follows the advice to do their own research. In this section, I will explain the kinds of advice for doing research that fringe theorists give and provide a graded account of what it means for research to be “your own.”
So far, I have mostly focused on the phrase ‘DYOR.’ This phrase, by itself, is a piece of advice about further inquiry. It instructs the audience to leave the question open and conduct further inquiry into the subject matter. It may also carry implicatures about which sources to trust and which not to trust, depending on the context. However, the phrase ‘DYOR’ is just the tip of the iceberg. Some fringe theorists advise audiences to use specific search engines (Thompson Reference Thompson2022), keywords, or search procedures. Some even provide collections of PDFs that they recommend using for further inquiry.Footnote 25 Whether fringe theorists simply tell audiences to conduct further inquiry or also give advice on how to proceed, the implication is that audiences should not conclude inquiry on the basis of the information provided by the speaker. As one creator says, “Remember to do your own research. I’m only here to plant seeds for you to grow” (Unnamed creator cited in Marwick et al. Reference Marwick, Pippert, Furl and Schnabel2025).
Looking at the types of advice listed above, one might wonder in what sense reading a bunch of provided PDFs counts as doing “your own” research. It will be useful to think of the independence of one’s research (in the colloquial sense) as coming in degrees. Our research is more independent (in the colloquial sense) the less we rely on the help of others to answer our question of inquiry. Our research is more independent from any particular other person the less we rely on that particular individual to answer our question of inquiry. At one extreme, the mythic Cartesian meditator is completely independent. At the other extreme, someone who uncritically believes whatever they are told is radically dependent. Between those two extremes, we may accept more or less help from others in answering our question. Given that independence is a graded notion, what counts as independent research will depend on your point of comparison.
One line of criticism of ‘DYOR’ is that independent research is an unattainable ideal (Buzzell and Rini Reference Buzzell and Rini2023). Expert scientists are incapable of doing purely independent research about questions like whether vaccines cause autism. Any research on this topic would be building on years of research conducted by others about what autism is and how to diagnose it. As scientific researchers become more specialized, they increasingly rely on others’ expertise whenever their research involves elements outside their hyperspecialized niche.Footnote 26 Certainly, laypeople, who are not allowed to conduct medical research studies and must rely on scientists for their data, are not capable of independent research into this topic. The most independent form of research that any layperson could realistically accomplish on a topic like whether vaccines cause autism would be accessing some raw data and conducting their own analysis. While critics are correct to point out that proponents of ‘DYOR’ fail to live up to an ideal of pure independence, it is not clear that the advice to DYOR is always recommending pure independence rather than simply recommending more independence than some relevant contrast.Footnote 27 After all, as mentioned in the introduction, while some fringe communities demand members do things like analyzing raw data, often people understand much less independent methods of inquiry (like internet research) as examples of doing their own research.
We can imagine a hierarchy of independence, running as follows:
-
• mythic Cartesian inquiry;
-
• gathering and analyzing your own raw data;
-
• analyzing others’ data;
-
• doing a comparative review of existing sources;
-
• selecting a single source to trust (on the basis of reasons);
-
• non-discriminating deference to any old speaker.
When people tell you to DYOR, they are instructing you to be more independent in this colloquial sense – higher up the hierarchy than whatever contrast case is relevant.
Is there something self-contradictory about a fringe theorist who advises their audience to do their own research yet provides a list of PDFs or search terms? Not necessarily. It depends on the contrast case. Certainly, an audience member who determines their own research procedure is more independent than an audience member who follows specific advice to read certain texts or search certain terms. After all, the audience member who follows specific advice is relying on the speaker to tell them how to conduct an inquiry into the question, while the audience member who determines their own procedure is not.
However, if the contrast case is forming a belief solely on the basis of the evidence provided by the fringe theorist, then using a specified research procedure is more independent. After all, an audience member who reads the PDFs relies less on the speaker than an audience member who relies on the speaker to summarize the PDFs. An audience member who runs a search using suggested search terms relies less on the speaker than an audience member who only sees the search results that the speaker highlights. Even when fringe theorists suggest specific search procedures, they are still encouraging their audience to do “their own” research rather than forming a judgement on the basis of the evidence the fringe theorist presented.
3. The value of independence
I will now argue that the same evidence for a fringe view will generally seem to have more evidential weight for an audience member when they discover the evidence via more (colloquially) independent research. Therefore, if your goal is to persuade your audience of a view, and you can get your audience to encounter evidence for your view via their own research, it benefits you to encourage them to conduct that research rather than presenting the evidence yourself.
At this juncture, it is important to disambiguate two senses in which research may be independent. So far, I have been using ‘independent’ in the colloquial sense where your research counts as independent from mine insofar as you do it without relying on me. There is a related but distinct way in which your research can be independent from mine that has to do with how well your research results predict my research results. The more certain an observer can be about what results I have from seeing your results, the less independent our results are in this non-colloquial sense. Readers will recognize the connection between this sense of independence and the notion of Bayesian independence.Footnote 28
This technical sense of independence is important because the evidential significance of a source depends partially on how independent it is from the sources we have already consulted. If our research results are not independent (because, for example, I am copying what you say), then once someone has learned your results, there’s nothing left to learn from my results. They can only gain further information from checking my results if our results are at least somewhat independent. Of course, independence is not enough for my results to have probative value. If I generate my results randomly, they will be independent from yours but also reveal nothing about the topic of our research. In order to have probative value, my results must be truth-conditionally independent from yours – in other words, given that you have a certain result, I must be more likely to have the same result if that result is true than if it is false.Footnote 29 If someone believes that my results are truth-conditionally independent from yours, then learning that I agree with you should raise their confidence in the truth of our results.
Now we are ready to relate the colloquial and the technical sense of independence. Notice that if my research is not very independent from yours in the colloquial sense because I am relying on you, my research results will not be very independent from yours in the technical sense. Since I am relying on your work, my results will likely be similar or the same as your results. As a rule of thumb, the more (colloquially) independent my research is from yours, the more probative value my results will have if I find results that agree with yours. After all, if I am not relying on your work, the best explanation if we return the same results will often be that we both successfully discovered the truth.
If the fringe theorist could get us to sit still and pay attention for long enough, they could simply show us whatever evidence there is for their view. Why tell us to do our own research rather than presenting the evidence themselves? If they show us the evidence themselves, we cannot expect our judgements on the basis of their evidence to be particularly (technically) independent from their testimony. Regardless of whether their view is true, we would expect them to show us evidence in its favor, and so if we only rely on evidence they present us, we are relatively likely to form the same judgement they did. On the other hand, if the fringe theorist instructs us to go do our own research, we may take our research to be more (colloquially) independent and our judgements to be more (technically) independent. The same evidence in favor of the fringe view will have more probative value if we discover it as part of our own research than it would if the fringe theorist presented it to us.
If we’re researching fringe theories, though, aren’t we likely to find disconfirming evidence?Footnote 30 By virtue of the fact that a view is fringe, there will be a lot of evidence against the view waiting for us to find it. Remember, the reason we afford the products of our own research extra evidential power is because we are more likely to find whatever disconfirming evidence exists when we do our own research than when we rely on what the fringe theorist directly communicates to us. If our research is more truth-conditionally independent, the chances that we find evidence against a fringe view are higher. The evidential boost we give to any confirmatory evidence we find may well be overwhelmed by all the counterevidence we find. Fringe theorists can only rest assured that we will not find evidence against their view if our research is not particularly truth-conditionally independent from their presentation. However, if our research is not particularly truth-conditionally independent from the theorist’s presentation, then we should afford the products of our research proportionately little evidential weight. The point is that, so long as we are correctly proportioning the extra weight we afford our own research to the degree of truth-conditional independence of our research, it is unclear how getting us to do our own research could help a fringe theorist lead us to accept their fringe view. The missing piece is that we are prone to overestimate how truth-conditionally independent our own research is by underestimating how much fringe theorists are able to steer our inquiry. When audiences conduct research under the illusion of independence, they will afford any confirmatory evidence they find more evidential weight than it merits. I will discuss how we may fall under the illusion of independence in the next section.
4. The illusion of independence
When someone advises their audience to do research, they are attempting to influence their audience’s inquiry. At a minimum, they are telling their audience to conduct further inquiry into a topic. Even when the only explicit instruction is to do more research, the speaker’s presentation may direct their audience’s research by shaping how their audience frames their research question. Moreover, as discussed in Section 2, fringe theorists often include more specific instruction about what methods of inquiry their audience should use. When audience members knowingly or unknowingly defer to the fringe theorist about how to inquire, the fringe theorists can direct their inquiry towards evidence that will confirm the fringe view. If an audience member recognizes that they are deferring to a speaker about how to conduct inquiry, they should lower their estimate of the independence of their findings. However, there are two ways that the illusion of independence can be maintained. First, if an audience member is unaware that they are deferring about how to conduct inquiry, they will not realize that they need to adjust their estimate of the independence of their findings accordingly. Second, even if an audience member is aware that they are deferring about how to conduct inquiry, they may not realize the extent to which accepting that method of inquiry renders their findings non-truth-conditionally independent. They may be under the illusion of independence.
Fringe theorists are able to bring their audience to encounter evidence that supports their view (and not evidence against their view) in large part because members of their audience, including those who won’t defer to the fringe theorist about the subject at hand, will defer to the fringe theorist about matters of inquiry.Footnote 31 It wouldn’t be unusual to defer to someone about one topic and not another if you have reason to expect them to be more trustworthy in a certain domain. However, it would be surprising if someone were untrustworthy about whether vaccines cause autism but trustworthy about whether and how to inquire into that question. Arguably, the reason we would not defer to fringe theorists about what to believe is precisely because we think they have flawed practices of inquiry. If we don’t trust them about what to believe on a topic, we shouldn’t trust them about how to inquire about that topic. There are costs, both practical and epistemic, to forming bad beliefs about whether and how to inquire. On the practical side, we might waste the opportunity to do something more worthwhile. On the epistemic side, framing our research question in a misleading manner or looking for evidence in all the wrong places can cause us to form false beliefs or fail to form true ones. Why, then, do people trust fringe theorists whom they wouldn’t trust about what to believe to tell them about how to inquire? Furthermore, if we follow their advice about how to inquire and find evidence that seems to support their view, we should not be particularly surprised or impressed. People should not afford this evidence extra weight if it is predictable, but my suggestion is that, at least in some cases, they do. Why might this evidence still hold extra weight for them?
I can think of two possible explanations for why people might defer about how to inquire and still believe their findings will be independent, each of which apply in some cases. The first explanation is that the people who defer about how to inquire are failing to recognize that they are deferring. People may fail to recognize that they are making a choice among competing options – options that will deliver different results. If they do not realize they are making a choice, they cannot recognize that they are deferring in making that choice. For example, a person may not realize that different, ostensibly synonymous search terms will pull up different results and therefore fail to recognize that they are deferring about which search terms to use.Footnote 32 Or, people may recognize that there is a choice to be made, but fail to recognize that they are deferring in their decision. So, for example, a person may realize that different search terms will pull up different results but not realize that they have absorbed the way a fringe theorist frames the question and used the theorist’s language in their search.
The second explanation is that people recognize that they are deferring about how to inquire but fail to recognize the danger of doing so. Such people may not recognize that inquiry can be misleading, or they may lack the humility to believe that they themselves could be misled. As a result, they believe their inquiry will only confirm the fringe view if that view is true. So, for example, someone may recognize that they’re deferring to the fringe theorist about which search terms to use but assume any apparently synonymous options would turn up evidentially equivalent results.Footnote 33 Such people are overly optimistic about inquiry in general. Others are only overly optimistic about their own inquiry. People who believe they cannot be misled by bad arguments will not see any danger in reading through a bunch of PDFs provided by the fringe theorist because they expect that they will form a belief on the basis of the readings only if they encounter a sound argument.Footnote 34 Both sorts fail to recognize that they run any risk by deferring. If you believe that any method of inquiry will point you to the truth, either because you believe that inquiry always leads to the truth or that your own inquiry always leads to the truth, then deferring about how to inquire will not strike you as jeopardizing the truth-conditional independence of your findings.
In general, this level of optimism about inquiry is unjustified. Empirically, people with low expertise in some domain tend to overestimate their abilities.Footnote 35 When laypeople attempt to do their own research, even free from the influence of a fringe theorist, they often make mistakes and fail to realize it (Levy Reference Levy2022). They may discover misleading evidence, engage in motivated reasoning, or exhibit confirmation bias all on their own.Footnote 36 However, there is still more reason for pessimism when lay researchers are deferring to an unreliable source about how to conduct their inquiry. Not all methods of inquiry are created equal – some methods of inquiry are particularly likely to lead lay researchers to find misleading evidence or draw mistaken conclusions. For example, using certain search terms will return exclusively low-quality sources. Using small datasets will not provide enough observations for any findings to be significant. Audience members who are unaware of these pitfalls may fall into such traps on their own. However, they may also be sent into such traps by fringe theorists offering unreliable guidance on how to conduct inquiry.
If we notice that we are riskily deferring to a speaker about how to inquire, we should judge the results of our inquiry to be less truth-conditionally independent from the speaker’s view. There are a few reasons for this. First, the speaker’s view may be based on the same research results if they used the same method of inquiry they suggested to us, so the speaker’s view is less likely to be independent from the results of this inquiry. Second, the speaker may have suggested this inquiry to us because they could foresee that it would lead us to evidence that appears to confirm their view, so the results of the inquiry are less likely to be independent from the speaker’s view.
Unfortunately, either because they do not realize they are deferring or because they do not realize the risk in doing so, people often judge their research results to be more independent than they are. When the audience is under the illusion that their research is more independent than it actually is, they will see their findings as having unwarranted evidential weight. This is the second way that instructing their audience to do their own research can help people with low credibility bring their audience to accept their view – it allows fringe theorists to put evidence for their view in front of the audience while maintaining the illusion of independence. By telling their audience to do their own research, fringe theorists encourage audience members to conduct further inquiry. The fringe theorists can then direct that inquiry towards confirmatory evidence. If the audience fails to recognize the extent to which the conclusion of their research was predetermined by the fringe theorist’s presentation, they may well afford that evidence a disproportionate amount of significance. The same body of evidence will be more persuasive if the audience takes themselves to have discovered it via relatively independent research than it would be if the fringe theorists presented it themselves.
5. Case studies
In this final section, I will share some examples of how this second rhetorical effect of ‘DYOR’ plays out. While virtue signaling is a fairly familiar and widely discussed phenomenon, fostering the illusion of independence may be more unfamiliar and harder to spot without examples.
So far, I have argued that it benefits fringe theorists to send their audience off to encounter evidence on their own when they can (1). direct their audience’s research to apparently confirmatory evidence while (2). maintaining the illusion that their audience’s findings are more independent (and thus more probative) than they actually are. In Section 4, I discussed why people may follow a fringe theorist’s guidance about when and how to conduct inquiry while still believing that their inquiry is truth-conditionally independent. However, one still might wonder whether people actually follow the advice dished out by fringe theorists. It is hard to say exactly how often people come to accept fringe theories after following the research advice of fringe theorists. There is some evidence to believe this is an effective strategy for converting people to fringe views. For one, fringe theorists who are self-consciously strategizing about how best to win converts highlight these techniques. Moreover, the accounts that fringe theorists give to retroactively explain how they came to accept a fringe view suggest that “independent” inquiry of the kind described here played a role.Footnote 37 In this section, I will discuss several case studies that I take to show how this rhetorical tactic works.
Wayfair and QAnon
In 2020, adherents of the QAnon conspiracy theory noticed that the online furniture store Wayfair was selling storage cabinets with human first names for very high prices.Footnote 38 They connected the names given to the cabinets to the first names of children who had been reported missing. They then theorized that Wayfair was trafficking missing children under the guise of selling storage solutions. Proponents of this theory instructed their audience to discover the truth for themselves by using the Russian search engine Yandex. The advice: take a product code from the Wayfair website, add the keywords ‘SRC USA’ (which are letters with special meaning to the human traffickers, according to the QAnon conspiracy theory), and hit search. Reliably, people who followed this advice would find softcore child pornography photos “confirming” the QAnon theory. The reason searchers would get the shocking results is that Imgsrc is a Russian website on which softcore child pornography proliferates. Searching any string of numbers with SRC is likely to turn up this kind of image on the search engine Yandex.
In this case, the conspiracy theorists are being fairly blatant about directing their audience’s inquiry. They specify what search engine to use and how to select search terms. However, the audience has at least one degree of freedom – they can pick any product code from the Wayfair website. By telling their audience to run the search themselves rather than reporting their own findings, the QAnon theorists change the source of the evidence to one that gives the illusion of independence. Because the audience members conduct the search themselves using any product code from the Wayfair website, they may expect that whether they find results that confirm the theory is at least somewhat truth-conditionally independent from the speaker’s own findings. Of course, in fact, Yandex will return “evidence” for the theory regardless of whether it is true or false. The respect in which the audience members are relying on themselves rather than the speaker (picking a product code) has no impact on what results they will find. Therefore, the apparent independence of their research is illusory.
Here, because the direction given is so blatant, the illusion of independence makes use of the audience’s ignorance about the unreliability of certain methods of inquiry. The trick is that there are predictable patterns in search results given certain keywords on Yandex. These predictable patterns allow QAnon theorists to direct their audience’s inquiry to misleading evidence. While it is unclear exactly how many people were convinced by this research technique, it seems that it was at least somewhat successful. In a support-group forum for those whose loved ones believe in QAnon, this tactic was discussed in terms that suggest it won some converts.Footnote 39 It’s worth noting that those who advise their audience to use this method of inquiry may not be aware that the evidence is misleading. They may have been misled themselves.
Crisis actors and data voids
In the aftermath of the Sandy Hook Elementary School shooting, a number of fringe theorists put forward the theory that the event never took place.Footnote 40 They said the event was staged, and the children and parents affected were “crisis actors” – actors pretending to be victims of an emergency event. At the time, there were few articles on the internet using the term ‘crisis actor.’ Very quickly, internet sources sharing conspiratorial views filled up the search results. Until the term ‘crisis actor’ attracted mainstream attention, searching that term would return a lot of conspiratorial content. Search terms that return mostly low-quality results are known as “data voids.” Until mainstream sources started talking about “crisis actors” in order to debunk the fringe theories using the term, ‘crisis actors’ was a data void.
Sometimes, fringe theorists would explicitly direct their audience to look up the phrase ‘crisis actor.’ For example, in 2014, Field McConnell and David Hawkins appeared on a podcast, saying that the Sandy Hook shooting was fake. McConnell said, “Let me just give your listeners a Google search of five items that will completely educate them on Sandy Hook.”Footnote 41 One of the terms he recommended was ‘crisis actor.’ Other times, fringe theorists direct their audience’s research more implicitly. In their choice of what words to highlight or emphasize, speakers mark specific terms as conceptually important to the topic, framing how audience members will conceptualize the topic of inquiry. Neologisms, like ‘crisis actor,’ provide a hook for audience curiosity. Even when the speaker does not tell their audience to search for those terms, they are directing the path of their audience’s inquiry.
Some conspiracists, hacks, and grifters intentionally adopt language they know will point their audience to sources that share their perspective. Others presumably use such language without this explicit purpose, simply because it’s the language with which they are accustomed to discussing this topic.
The trick here is using language that the audience does not realize is exclusively or predominantly used by people who share the fringe theory. Audience members may fail to realize that they are deferring (especially when the direction is implicit) or may fail to realize that by deferring, their search results are less truth-conditionally independent. In fact, by deferring in their choice of search terms, the results of their search are likely to confirm the fringe theory, whether it is true or false. If the audience assumes their results are more independent from the fringe theorist’s testimony than they are, they may well afford those results extra evidential weight.
I will now share one final example of this phenomenon that demonstrates the danger of mainstream advice to DYOR (like, for example, off-hand advice to fact-check online sources). Instead of occurring “in the wild,” this example comes from an experiment about the impact of researching fake news.
Fact-checking fake news
From 2019 to 2021, a group of researchers ran a series of studies about how researching false news articles impacts participants’ ratings of the articles’ veracity.Footnote 42 In these studies, researchers showed participants a number of true and false news articles and asked participants to rate whether they were true. For the experimental condition, they instructed participants to conduct Internet research into the article to see if it was true. They found that participants in the experimental condition were more likely to rate the article as true than those in the control, even when the article was false and even when participants initially rated the article as false.
In this study, the researchers instructed the participants to seek out further evidence that bears on the views expressed in the article. In one iteration of the study, participants’ research processes were tracked. As might be expected, participants who became more convinced that a false story was true had found unreliable sources when they conducted their internet search. One strategy that participants adopted was to use the title of the article as a search term. When participants used this strategy to research false new articles, they were especially likely to find unreliable sources. The reason is that the false new articles’ titles often included terms that were data voids at the time (like ‘engineered famine’).
There are some important differences between what happened in this case study and the phenomena of interest, and yet we can still learn some interesting lessons, both about fringe advice to DYOR and about more mainstream advice, like the advice to fact-check online sources.
Unlike when fringe theorists advise audience members to do their own research, in this case, the researchers instructed participants to do their own research as part of participating in the study. Keeping in mind the different context, it is still interesting to see what happened when participants conducted their own research. Even though the authors of the false articles were not providing explicit instruction about whether or how to conduct further research into their topic, the authors’ choices about framing, like how to title their article, still influenced the course of (some) participants’ research. By using terms that point to data voids and highlighting this terminology in the title, the authors of the article (intentionally or unintentionally) directed the participants’ inquiry toward low-quality sources that apparently confirm the article. If participants were aware that using the title as a search term would lead to results that are not particularly truth-conditionally independent from the initial article, they ought not to have used the title as a search term, or they should not have taken the results of that search to have significant evidential value. Yet, apparently, they did.Footnote 43
One of the striking features of this case study is that the ones issuing explicit instructions about inquiry, the researchers, did not intend for that inquiry to lead participants towards fringe sources. Yet, when participants applied that generic advice to researching fringe views, they tended to find fringe sources. Again, there are differences between the context of this study and the ordinary contexts in which mainstream speakers urge their audiences to fact-check online sources. However, the study provides at least some evidence that the generic advice to conduct further inquiry into online sources may have unintended consequences.Footnote 44 If people who follow the advice to fact-check online claims use similar approaches to the participants in the study, then when they encounter fringe claims online, their research may well lead them to be more confident in the fringe view than they were before looking it up.
If we are going to advise people to do further inquiry into stories they see online, we must also give advice about what form that inquiry should take. In particular, we should give guidance about how to pick appropriate search terms and how to interpret search engine results.Footnote 45 People should come up with search terms themselves when possible and experiment with several search terms that would ostensibly pull up the same information. Search engines are not necessarily answering the question you had in your mind (even when the AI summaries return what sounds like a natural language answer to that question), and the articles that show up on the first page of results are not necessarily the only or best sources for your inquiry. Unless the advice to fact-check sources is accompanied by instructions about how to do so, the direction of listeners’ research may be influenced by the very source they are attempting to fact-check, or so the study discussed above suggests. While the people running the study would follow up with participants explaining which articles were false and why, we can’t always do the same for those we urge to research Internet claims.
All that said, sometimes a mainstream speaker may find themselves in a context where they can advise audience members on how to responsibly do their own research that will turn up non-misleading, truth-conditionally independent results. Doing so may get evidence for their mainstream view in front of audience members in a way that legitimately has more evidential weight than it would if the mainstream speaker were to share the evidence themselves. If mainstream speakers care that their audience ends up with a fully justified credence in the mainstream view, they must take care not to allow audience members to harbor an illusion of more truth-conditional independence between the speaker and the layperson’s inquiry than actually exists. In those cases where mainstream speakers can do so, the advice to do your own research may be legitimately employed. These cases will likely be rare, but they may be especially valuable for reaching people for whom mainstream speakers lack credibility.
6. Conclusion
It may seem pointless yet innocent for fringe theorists to encourage their audience to do their own research rather than deferring to the theorist’s presentation about the subject at hand. However, since fringe theorists can (intentionally or unintentionally) direct their audience’s research such that they find evidence supporting the speaker’s view, this strategy is less self-defeating than it may appear. In fact, for speakers who start with a deficit of credibility, instructing their audience not to defer can raise their credibility by signaling their epistemic virtue and may lead their audience to be more persuaded than they would have been if the speaker simply made their presentation. The audience may be more persuaded if encouraged to do their own research insofar as they will interpret the evidence they found “on their own” to provide independent support. In the end, fringe theorists encouraging their audience to do their own research rather than deferring to their presentation is neither pointless nor innocent.Footnote 46
Competing interests
The author declares none.
Open access
