Search ideas

Say someone said "I had a dream that {insert something true}" or "god told me that {insert something true}," what is the source of knowledge here?

This has to take time into context. At one point, a belief in god was all that we had. We didn't have hard to vary explanations. As such, a person might have a belief in god as the only worldview currently. So it isn't irrational for that person, or people back in the days, to believe in god.

Dreams can be a source of knowledge. But dreams aren't always reasonable. Sometimes, dreams are lies.

In that statement, I am looking at reason as a mode of criticism. You might get ideas and potentially knowledge from all sources and reason tests weather they are right or not.

And if I understand you right, what you're saying is if an idea isn't from 'reason' than how can we criticize it using reason. But we can and do all the time. Religion is irrational, but we criticize it and take what is good from it and discard the rest.

What Does “Battle Tested” Mean?

One of @edwin-de-wit’s ideas recently got the blue label that says “battle tested” – well done, Edwin! – so he asked me what it means.

It means that the idea has at least three criticisms, all of which have been addressed.

The label is awarded automatically. It’s a tentative indicator of quality. Battle-tested ideas generally contain more knowledge than non-battle-tested ones.

When there are two conflicting ideas, each with no outstanding criticisms, go with the (more) battle-tested one. This methodology maps onto Popper’s notion of a critical preference.

The label is not an indicator of an idea’s future success, nor should it be considered a justification of an idea.

You can see all battle-tested ideas currently on Veritula on this page. Those are all the best, most knowledge-dense ideas on this site.

Recursive Epistemology

Veritula implements a recursive epistemology. For a criticism to be outstanding, it can’t have any outstanding criticisms itself, and so on, in a deeply nested fashion.

def criticized? idea
  outstanding_criticisms(idea).any?
end

def outstanding_criticisms idea
  criticisms(idea).filter { |c| outstanding_criticisms(c).none? }
end

def criticisms idea
  children(idea).filter(&:criticism?)
end

This approach is different from non-recursive epistemologies, which handle criticisms differently. For example, they might not consider deeply nested criticisms when determining whether an idea is currently criticized.

How Does Veritula Work?

Veritula (Latin for ‘a bit of truth’) provides an objective, partly automated way to tentatively determine whether a given idea is true or false.

It follows Karl Popper’s epistemology, which says that ideas are assumed true until refuted. This approach leaves us free to make bold conjectures and use the full arsenal at our disposal to criticize these conjectures in order to correct errors and seek truth. It’s a creative and critical approach.

Veritula is a programmatic implementation of Popper’s epistemology.

Consider an idea I:

              I

Since it has no criticisms, I is considered unproblematic. It is rational to adopt it, tentatively consider it true, and act in accordance with it. Conversely, it would generally be irrational to reject it, consider it false, or act counter to it.

Next, someone submits a criticism C1:

              I
              |
              C1

The idea I is now considered problematic so long as criticism C1 is not addressed. How do you address it? You can revise I so that C1 doesn’t apply anymore, which restores the previous state with just the standalone I (now called I2 to indicate the revision):

                   Revise
              I ------------> I2
              |
              C1

To track changes, Veritula offers beautiful diffing and version control for ideas.

If you cannot think of a way to revise I, you can counter-criticize C1, thereby neutralizing it:

              I
              |
              C1
              |
              C2

Now, I is considered unproblematic again, since C1 is problematic and thus can’t be a decisive criticism anymore.

If you can think of neither a revision of I nor counter-criticism to C1, your only option is to accept that I has been (tentatively) defeated. You should therefore abandon it, which means: stop acting in accordance with it, considering it to be true, etc.

Since there can be many criticisms (which are also just ideas) and deeply nested counter-criticisms, the result is a tree structure. For example, as a discussion progresses, its tree might look like this:

              I
           /  |  \
         C11 C12 C13
         / \       \
       C21 C22     C23
                   / \
                 C31 C32

In this tree, I is considered problematic. Although C11 has been neutralized by C21 and C22, C12 still needs to be addressed. In addition, C23 would have neutralized C13, but C31 and C32 make C23 problematic, so C13 makes I problematic as well.

But you don’t need to keep track of these relationships manually. Veritula marks ideas accordingly, automatically.

Because decision-making is a special case of, ie follows the same logic as, truth-seeking, such trees can be used for decision-making, too. When you’re planning your next move but can’t decide on a city, say, Veritula helps you criticize your ideas and make a decision. Again, it’s rational to act in accordance with ideas that have no outstanding criticisms.

All ideas, including criticisms, should be formulated as concisely as possible, and separate ideas should be submitted separately, even if they’re related. Otherwise, you run the risk of receiving ‘bulk’ criticisms, where a single criticism seems to apply to more content than it actually does.

Again, criticisms are also just ideas, so the same is true for criticisms. Submitting each criticism separately has the benefit of requiring the proponent of an idea to address each criticism individually, not in bulk. If he fails to address even a single criticism, the idea remains problematic and should be rejected.

The more you discuss a given topic, the deeper and wider the tree grows. Some criticisms can apply to multiple ideas in the tree, but that needs to be made explicit by submitting them repeatedly.

Ideas that are neither criticisms nor top-level conjectures – eg follow-up questions or neutral comments – are considered ancillary ideas. Unlike criticisms, they do not invert their respective parent’s truth status. They are neutral.

One of the main benefits of Veritula is that the truth status of any idea in a discussion can be seen at a glance. If you are new to a much-discussed topic, the rational course of action is to adopt the displayed truth status of the ideas involved: if they are marked problematic, reject them; if they are not, adopt them.

Therefore, Veritula acts as a dictionary for ideas.

One of the problems of our age is that the same discussions are had over and over again, sometimes by the same people. Part of the reason is widespread irrationality, expressed in the unwillingness to change one’s mind; another is that it’s simply difficult to remember or know what’s true and what isn’t. Discussion trees can get complex, so people shouldn’t blindly trust their judgment of whether some idea is true or problematic, whether nested criticisms have been neutralized or not. Going off of memory is too error prone.

Veritula solves this problem: it makes discussion trees explicit so you don’t have to remember each idea and its relation to other ideas. Veritula therefore also enables you to hold irrational people accountable: if an idea has outstanding criticisms, the rational approach is to either abandon it or to save it by revising it or addressing the outstanding criticisms.

Many people don’t like to concede an argument. But with Veritula, no concessions are necessary. The site just shows you who’s right.

Using Veritula, we may discover a bit of truth.

good idea!

For example, I had to manually notify Edwin in #1811 of a revision I had made to address a criticism of his. Without this notification, he might miss the revision. If he disagrees that the revision addresses his criticism, that’s a potential error that might not get corrected.

When you revise an idea to address a criticism, its author should get a notification so they get a chance to verify that the revision really does address the criticism.

@edwin-de-wit In light of the potential shortcomings of Deutsch’s definitions of knowledge, I’ve revised my criticism, resulting in #1806. See if you want to counter-criticize it.

Yeah

But this thread, about my English possibly being a problem, doesn’t seem like a relevant or substantive criticism.

As I’ve pointed out previously, I wouldn’t try to assign strengths (or ‘substantiveness’) to arguments.

Any criticism no matter how small destroys its target decisively if unaddressed. Whether or not its decisive is determined by whether or not there are any counter-criticisms, not by assigning some strength score (a remnant of justificationism). A criticism is decisive as long as there are no counter-criticisms. In the absence of counter-criticisms, how could it not be decisive?

In the current situation, this epistemology is actually to your benefit because, if some idea (such as #1731) is off topic, simply pointing this out in a criticism completely neutralizes the idea you deem off topic.

If a criticism really is tiny (or ‘weak’), it’s easier to just correct the error it points to than to counter-criticize. (For example, it’s usually quicker to fix a typo than to argue about the merits of pointing out typos.)

This is how Veritula is built. If you have an epistemological disagreement about its functionality or want to continue this broader epistemological discussion, submit an idea or criticism in ‘How Does Veritula Work?’.

Since you asked, I suggest you do both of the following:

1. Submit a criticism of #1731, suggesting that your English is off topic because the carelessness you suggest caused your typos had no effect on your choice of new terminology (‘Statements’ etc).
2. Since you agree that #1738 is not a criticism, revise it to unmark it as a criticism. At the bottom of the revision form, uncheck #1799 to indicate that it does not apply anymore.

Superseded by #1806. This comment was generated automatically.

Another problem with the term ‘statement’ is that not every statement encodes knowledge. Only some statements do.

For example, trivial or tautological statements aren’t knowledge, neither in the Popperian nor common sense of the word.

You make a good point. Maybe the definition ‘information with causal power’ on its own isn’t very good since virtually any statement can have causal power yet presumably even Deutsch would agree that merely saying ‘hi’ isn’t knowledge even though it can have causal power (eg prompting someone to say ‘hi’ back).

Gotcha! Did my most recent edit now address the criticism that Joy isn't signaling an unaddressed conflict?

Superseded by #1802. This comment was generated automatically.

I think it does imply a conflict. Every emotional sensation—including urges—arises to provide feedback to our consciousness about how a particular problem (in the Popperian sense: two or more incompatible theories in conflict) is or isn’t progressing.

For example, consider hunger. One theory (Drive A) is that we don’t want to be hungry, while another signals that we are hungry (from ephemeral sense data (which could itself be viewed as a Drive, though that’s not important here)). The conflict between these theories produces the urge — in this case, the sensation of hunger.

I explain these conflicts in more detail, with further examples of Drives, Intuitions, and Statements, in this video:

https://www.youtube.com/watch?v=uEcR_0GbzRE

Addition 01-09-2025:
In the case of hunger, the sensation was signaling an unaddressed problem, but as you correctly pointed out, not all emotions signal unaddressed problems. Emotions are a feedback mechanism that can reflect different stages of problem solving. For instance, joy may signal a resolved problem, and impatience might signal frustration with an ongoing one. Likewise, anxiety can serve as an early warning of potential obstacles ahead, while relief marks the successful removal of a previously pressing issue.

Yeah, it doesn’t feel like a real criticism. I’m just trying to figure out the right way to resolve this thread. You’ve raised other criticisms focused on the content of my explanations—those make sense to keep open. But this thread, about my English possibly being a problem, doesn’t seem like a relevant or substantive criticism. I've claimed that my inaccuracies come more from carelessness than from a lack of comprehension of the language, and that doesn’t feel like a criticism of the ideas we’re discussing. So what should we do with this thread?

Interesting, I hadn’t thought of that angle before. I’ve always taken a fairly broad view of “information with causal power,” assuming that any explicit statement from a human mind qualifies. Even the simple remark “Nice weather we’re having” can have causal power—it might prompt the listener to respond, or push the speaker to continue if the comment goes unacknowledged. In that sense, almost any statement can be read and potentially inspire another universal explainer. Even when fed into an LLM, the statement can still be parsed and worked with. In contrast, mere “information” in the form of gibberish, a made-up language, or a nonsensical string of random words would not be parsable, and therefore would not exert causal power on the parser.

I also recall Deutsch often saying that knowledge is information that tends to remain instantiated once it appears. I always understood that as a form of causal power, rather than as a separate criterion. I’m not sure he has ever been fully explicit on this point. But if he does mean it as a strict demarcation—that knowledge is only what causes itself to persist—then I’d agree with your criticism.

Cool, appreciate it. Since you agree and plan to be more precise, should this really be marked as a criticism?

By the way, you don’t need to put disclaimers like “Addition 01-09-2025”. The versioning system records and displays all that information automatically :)

Superseded by #1796. This comment was generated automatically.

[J]oy may signal a resolved problem […]

But then the conflict is gone. So I don’t think revision #1741 addresses #1730.

To be clear, when I asked about the conflict behind joy, I meant ongoing conflict.

Your addition seems to agree with my criticism, not address it.