It feels like you can’t believe anything anymore. Any heuristic we previously had to assess the trustworthiness of information has failed on us. There was a time where facts were more likely to be true, simply because they were written down, but that ended with the internet. There was a time where you could assume something was true just because you’d hear it repeated independently several times, but that ended with social media. There was a time where photographic evidence indicated truthfulness but that ended with photoshop. And finally, there was a time where the impression of high production value correlated with high factual accuracy, but that ended with generative AI. What we’re left with is a world where no information can be fully trusted.
What you can still do is trust people. It’s great if you know people who will put in the effort to rigorously research their claims to build a reputation of trustworthiness. They can even form trust networks as they can vouch for each other. And while I believe that this will become a lot more important in our new low-trust world, this is not what I want to talk about. Instead, I want to ask the question: is low-trust information actually worthless? Should you ignore sources just because they might be wrong? Or is there even merit in actively seeking out information that may be inaccurate?
To start this discussion, it probably helps reframing it in terms of eidon, as it allows us to define it more concisely: is it possible to improve the usefulness of your eidon by integrating non-trustworthy information? As a reminder, your eidon is all about prediction. So another way to look at this question might be: can you make useful predictions by considering non-trustworthy information? Note that usefulness is situational, so depending on the individual circumstances and personal values it might mean different things. However, in the vast majority of cases it is simply synonymous with accuracy. Since this post is not about religious belief or moral values, I’ll make this simplification here and say that predictions are most useful when they are accurate.
So how could you make accurate predictions if you rely on information that is not necessarily correct? Well, let’s start with the easiest case where you have ways to check whether the information is correct. You might consider that cheating because it seems like you didn’t need the information in the first place. And that is true because you had an alternative way to arrive at correct information which is guessing. You could have simply guessed something and then validated if it was true and if it wasn’t guessed again. But that is just an extreme case of the same approach, starting with maximally untrustable information. It highlights another big component of this question which is efficiency. Information from a bad source that at least claims to be correct might be a better initial guess than what you could come up with so you’ll arrive at the correct answer faster and with less effort.
But that was just the easy case. There are a lot of questions where you can’t easily validate the answer. In that case, your prediction could never be more accurate than the information you rely on. The question is, can it be more accurate than your prediction without the information? This depends on the reason the information might be wrong. We can group these reasons into two main categories: imprecise methods and adversity. The former case includes independent but underfunded news outlets, uncontroversial social media posts and, most prominently, LLMs. Whether or not you should use this information simply boils down to comparing measures of precision. If the likelihood of your source being right is higher than the likelihood of you guessing, it is useful to you. And while it is sometimes difficult to estimate this likelihood, the far more problematic case is the latter category of reasons.
Unfortunately, adversity is a common issue when it comes to false information. Deliberate misinformation is common in political social media posts, biased news outlets and conspiracy theories. The main difficulty is that it is designed to be difficult to detect. Needless to say, you shouldn’t incorporate such information into your eidon, but that is easier said than done. There are a few red flags to watch out for, like when it sounds too easy, too good to be true or too convenient. I don’t want to go into too much detail here as this is a whole different topic. The gist is that misinformation is a valid reason to completely dismiss a source of information, no matter how tempting it might be to believe it.
Getting back to the original question, there are valid ways to incorporate information from less trustworthy sources into your eidon to improve the accuracy of its predictions and thus its usefulness. However, you have to stay aware of it and ideally remember where your information came from or at least that it is uncertain. This takes more effort than incorporating information from a trustworthy source, but it may still be worth it if no trustworthy source exists as, unfortunately, is often the case these days. Importantly, when communicating the information to others you should probably communicate the uncertainty as well. This can be very problematic because uncertainty is unattractive. People who speak with confidence are more desirable to others, even if they are wrong sometimes. It’s the main reason there are so many imprecise sources out there but it’s tempting to be overconfident. And that is fine, as long as you can live with being perceived as an imprecise source. It’s a decision you’ll have to make for yourself.
This post is already getting quite long, but there are other benefits of listening to untrustworthy sources, even adversarial ones, as long as you don’t incorporate them in your eidon as facts. The main one being that they might be facts in other people’s eidons and it can be really useful to make predictions about what other people’s eidons might predict. Another one is that misinformation can reveal a lot about the motifs of those spreading it. And finally, some facts that are almost certainly wrong might still be fun. While it’s important to try and make your eidon as accurate as possible, that process is also quite boring and you shouldn’t underestimate the power of having a little fun along the way.
