Paths Forward to Correct Errors

Table of Contents

The Problem

If I’m mistaken about this idea, how will I find out?

A Path Forward is an answer to this question. It’s a way to make progress – a way to find out about and correct a mistake. You should ask this question often because without a good answer you’re at risk of being mistaken and staying mistaken in the long term.

A second question to ask is: What will be the negative consequences if I’m mistaken about this idea? In other words, what’s at risk?

Another way to think about Paths Forward is, “What is the most reasonable series of actions by which a smart, correct outsider can correct me about this, even if their correction sounds intuitively wrong to me?” What would it take for someone without social status, credentials or any social network to correct me? What if the correction is counter-intuitive and requires learning new background knowledge to understand? What if it initially sounds dumb to me and doesn’t appear to be worth my time – if they’re actually right, what steps would work to correct me despite my initial negative reaction? (You can write down Paths Forward – e.g. series of actions that would work well to correct you – and share them so that people know the best ways to try to correct you and so they can give feedback on how you could better enable error correction.)

Paths Forward is most important for public intellectuals (people trying to truth-seek and make progress, and who participate in public discourse, whether they have any credentials or not), and lots of my writing about it is targeted at them. It has relevance to everyone since it’s fundamentally related to how epistemology works.

Paths Forward thinking builds on the fallibilist, evolutionary epistemology of Critical Rationalism and Critical Fallibilism. Learning works by generating ideas and correcting mistakes, so Paths Forward are a fundamental part of learning. We can’t prove, support or justify our ideas; the best you can say about an idea is that it hasn’t been refuted so far. And we’re fallible – we commonly make mistakes and can never get guarantees that an idea isn’t mistaken – so we should always be taking into account the possibility of mistakes by having some plan so that we don’t get stuck. What are you doing about your fallibility, both generally and for specific ideas?

Creating and putting into words an approach to finding and correcting errors puts us in a better position to analyze and improve it, and to consistently use it.

Reasonable people already try to find and correct their mistaken ideas. Being imperfect, they inevitably miss some mistakes, and make some new mistakes when attempting corrections. Try to imagine how many mistakes people from a million years in the future would see us making.

So smart people commonly have an attitude like: do your best to seek the truth, try to use good judgement, critically consider your own ideas, and learn about the methods of reason. This is too vague and doesn’t specify what actions to take.

And there’s a big mistake most smart people are making! Consider this modified question:

If I’m mistaken about this idea, and someone else understands the mistake and is willing to share a better idea, how will I find out?

If you don’t think of something, that’s fine – you can’t figure everything out. And if no one thinks of something, that’s fine – we all missed it, that’s bound to happen sometimes. But what if someone does figure an issue out, and would be happy to tell you, and you stay mistaken anyway? That’s an avoidable failure.

If people could stop missing out on great ideas which are already known, that’d be a huge improvement. This is hard because there are a ton of ideas in the world (too many to read through them all, let alone discuss them).

Paths Forward explains how people miss opportunities to learn from others, and how to fix this problem. It proposes a better way to organize learning and knowledge, including specific actions to do. The goals are to better enable you to find and correct your mistakes and to better enable other people to help you.

The Solution

As you learn, gather written material which you can refer to which expresses what you believe to your satisfaction (you will endorse it and take responsibility for its correctness as if you wrote it yourself). To the extent you either can’t find adequate material, or you create new ideas, then write it yourself.

Don’t just learn things. Get them in writing. (In theory, it’s possible to use other mediums, like a video of a lecture, as long as it can be made publicly available on an ongoing basis so that discussions can refer to it. But writing is pretty dominant because it’s the best format. Important video and audio should be transcribed.)

Less than half of this writing should be positive explanations of what you believe and why. You also need answers to potential questions and criticisms, and you need criticisms of contradictory rival positions.

If you build your knowledge this way, it’s easy to answer large numbers of intellectual inquiries: refer people to the pre-existing writing which addresses their question, criticism, or rival idea. Unless someone says something new to you, you can respond to any discussion point with a reference.

This enables you to let the public contact you. You can engage with anyone who thinks you’ve made a mistake because it’ll be fast unless they say something new or important. For each of their questions, criticisms and alternative ideas, you already have writing to address it (or else it’s worth some attention)! New ideas will take more thought to address, but can come in at a manageable rate and be worth the effort.


What is a reference? Typically it’s an internet link or a citation to a book or paper. It can be any clear way to point someone to information which is publicly available in a stable, longterm way. Any time you refer someone to information, it should be treated the same as if you wrote it yourself, today. Think of using a reference as copy/pasting the entire text of the reference as your discussion reply. If that would be appropriate, great. If not, don’t use it, or refer to only a section of it. You can also refer to something but add some qualifiers, caveats, explanations of which parts you agree and disagree with, etc.

What if you’re so popular that you get too many inquiries to give even short responses like references to pre-written answers? Then you should be popular enough to have a discussion forum when your fans respond to inquiries with references (and you or your fans can put together FAQs and other documents to make it easy to find the references you use to answer common inquiries). With that much success, you should also be able to earn the money (or get donations) to hire an assistant to answer inquiries for you (as with fan help, they can answer the easy ones and pass the interesting ones on to you).

What about the problem of bad references? People often try to refer someone to an entire book instead of just the relevant part. Worse, sometimes the whole book is irrelevant. To solve this, create a library of good references (you can do this publicly on a blog, or keep it as notes that you can copy/paste). The references can include specific sections (possibly from multiple sources) along with a few sentences of explanation summarizing how the references are relevant to the issue. If you receive a reference you suspect is inadequate, ask for a more specific reference and a couple sentences stating what you will find at the reference and how it matters to the issue.

What about a succession of bad references? People sometimes give you a bad argument, you refute it, and they give you another bad argument, and they repeat this forever. This can happen regardless of whether the arguments are freshly written or are done via reference. Handle this by pointing out and criticizing the pattern of bad arguments.

What about a succession of bad questions or criticisms? Sometimes people keep asking about the same issue in different ways, or try to criticize the same point with different words. What should you do when people get repetitive? Speak to the theme, pattern or principle involved. E.g. they keep talking about concrete examples, but you recognize they’re making the same conceptual mistake every time. Reply about the conceptual mistake instead of getting caught up in the details. A rule of thumb is to address specifics three times then cover the more general issue. Going through some specific examples helps people understand it better, but doesn’t need to be done endlessly if you can recognize some kind of repeating issue to address in a more general purpose way. In the future, you can refer to specific examples you already covered in the past and then move on to referring to an explanation of the general issue, rather than doing three more examples.

How do I organize my references? Have a relatively small number of pieces of writing which explain general principles and address big categories of issues. Those are the hard and important parts. In addition to that, as you participate in discussion you can build up bridging material.

Bridging material is a short explanation that says how a general principle answers a specific issue and then gives a reference to an explanation of the principle. This is generally pretty quick and easy to create because it’s typically around a paragraph long.

Bridging material is what you usually want to refer people to (or create if you don’t have it yet) so they can get an answer tailored to their question or criticism. This way, you have lots of customized answers to many different issues, but they’re cheap to make because they consist of a small amount of unique writing plus one or more references. If you create material like this each time you discuss, you’ll quickly go from just having some main ideas covered to also covering many specific inquiries in a reusable way. Before long, you’ll build up great, specific, customized answers to most inquiries, so discussion won’t take much effort.

Intellectual Litter

There’s a common way people misunderstand Paths Forward. They think Paths Forward is the intellectual equivalent of how to pick up and dispose of every piece of litter you see, anywhere in the world. In other words, Paths Forward says how to spend all your time answering stupid questions from anyone who wants to waste your time. In short, people are worried that doing Paths Forward is way more effort than it’s worth. (Thanks to Josh Jordan for telling me that objection with the litter analogy.)

It’s important to allocate resources to error correction. That should include both looking for our own errors and enabling corrections from others. This shouldn’t use all of our resources. We should also allocate resources for making forward progress.

The resources we allocate for error correction from others should be used efficiently. How can we best enable error correction without it being too much work? That’s one of the major topics Paths Forward discusses. I’ve written about methods to lower the effort required, including references, proxies and argument reuse.

Most public intellectuals today don’t have Paths Forward. There’s no reasonable way for people outside of their social circle to correct them when they’re wrong. So first we need to convince people that Paths Forward matter. But after that, yes, making them more efficient and convenient is also really important. There’s certainly room for improvement on my initial answers for to how to do time-and-effort-efficient Paths Forward.

Keep in mind fallibility. People can misidentify what is litter. So there needs to be material specifying what categories of litter you’ve identified, the properties of the litter, how and why you know they’re bad, what methods you use to detect them, why you think those methods are accurate, ways this could go wrong, contingencies if you make a mistake, etc. Instead of assuming a conclusion about what’s litter – with no way to be corrected if you’re wrong – address it in a Paths Forward compatible way.

If you rely on intuitive assumptions about which things are intellectual litter and why, there will be bias and error – and no error correcting process. I think people need to learn how to quickly, cheaply argue with bad ideas. If it’s so dumb, it shouldn’t take you much effort to deal with it. If bad arguments are repetitive, then someone should address the patterns. If someone has already done that, you can refer to their work. If no one (that you know of) has ever written down why something is bad in an adequate way, then you shouldn’t assume it’s bad and you shouldn’t expect to throw it in the litter basket in a few seconds. If something doesn’t fit any known category of bad ideas, then it ought to be addressed. If it does fit a known category, you should say what category and link to arguments, rather than be silent, so that a rebuttal is possible and also so that the other guy can learn about his error if he’s wrong. Paths Forward doesn’t just help you avoid being and staying wrong; it also provides ways for others to learn from you and be referred to writing that can teach and persuade them.

The truth is not obvious. Don’t bet your career on your intuitions. Don’t approach thinking so that, if you’re biased about what looks like intellectual litter to you, you’ll permanently reject a bunch of great ideas. It’s OK to be wrong about issues where no one knows better. We can’t be perfect. But don’t ensure you stay wrong about errors that people have figured out and are willing to explain to you.

More Details

If you’re interested in trying to do Paths Forward yourself, you’ll want to study it more. Check out my Paths Forward Summary and the links to more resources at the end.