How Does Veritula Work?
Showing only those parts of the discussion which lead to #2214.
See full discussion·See most recent related ideasLog in or sign up to participate in this discussion.
With an account, you can revise, criticize, and comment on ideas.@dennis-hackethal you have regularly pointed out to me that it’s a mistake to assign strengths or weaknesses to arguments—for example, in #1809 and #1927. I’d love to get to the bottom of that.
On one hand, I see what you mean. A criticism can either be counter-criticized or it can be an unresolved error. If it’s a bad criticism, you quickly counter it—say, by clarifying why it’s irrelevant—and move on.
I also see why talking about a “gradient” or comparative strength between arguments is problematic: there’s no objective criterion to measure them against. We can only say one theory is better than another when both attempt to explain the same phenomenon—then we can evaluate them using properties such as hard-to-varyness and other criteria Deutsch describes. (We can get into that if you’d like, though I don’t think that’s our main disagreement.). But this comparison doesn’t apply when we’re dealing with very different criticisms of a single idea, because there’s no common standard to measure them against. Comparing their “strength” becomes arbitrary.
However, I still think there are good and bad criticisms, just as there are good and bad explanations (following Deutsch’s distinction: for instance, bad explanations are easy to vary or point to authorities to justify themselves rather than offering a hard-to-vary account of how and why something works). While I could simply counter-criticize bad criticisms and move on, there’s also the matter of efficiency and opportunity cost: I don’t want to waste time repeatedly countering poor criticisms, or worse, get stuck in circular debates with people who don’t recognize that some arguments aren’t good criticisms at all. I’d rather focus my attention on good criticisms.
To clarify what I mean, here’s an excerpt from my book:
The most important principle to remember while criticizing is: Criticize, don’t defend or attack. Good explanations invite criticism of their intrinsic content—whether the explanation itself works, solves the problem, and avoids worse side effects. Bad explanations, by contrast, deflect criticism onto irrelevant, extrinsic properties such as authority or track record—e.g., “this is the method that successful company X uses,” “I believe strongly in this approach,” or “it’s coming from person Y, so it’s worthless.”
That kind of “criticism” isn’t real criticism at all. It’s just attacking or defending. And when we play that game, the explanation itself stays untouched and stagnant. The idea doesn’t get scrutinized or improved—it only gets shielded or dismissed for irrelevant reasons.
That being said, I agree with the points you brought up in #2061, namely that there can be no positive arguments for an idea. By extension, if Veritula would require a specific format or mode of criticism, we’d fall into the very error Popper warns about with the Myth of the Framework—the mistaken belief that criticism requires a shared framework or language. So, Veritula should functionally remain as it is. At most, you might consider adding guidelines on what constitutes good versus poor criticism, so that critics can improve their skills. But I agree: the person who created the idea should remain solely responsible for addressing the criticisms they receive, not dismissing them as “bad” and moving on.
Tagging @bart-vanderhaegen because he and I have discussed this at length—in fact, I got the defending/attacking framing from him.
Pasting #2079 here as it’s since been hidden in a resolved child thread and should have applied directly to #2074 in the first place.
My current view is that the only meaningful dichotomy is good vs. bad.
You say yourself in #2071 that one should “always avoid positive arguments.” Calling a theory “good” would be a positive argument.
As I say in #2065, Popperian epistemology has no room for ‘good’ or any other justification. I’m not aware that anyone has successfully proposed a way to measure the ‘hard-to-varyiness’ of theories anyway. We can criticize theories for being arbitrary (which is another word for ‘easy to vary’). That’d be fine. But Popper wouldn’t give them points for not being arbitrary. And arbitrariness isn’t the only type of criticism a theory might receive anyway.
If we follow Popper and get rid of justification, we can’t use ‘good vs bad’ because we can’t use ‘good’. The only dichotomy left standing is ‘has some bad’ vs ‘has no bad’. Another word for ‘pointing out some bad’ is ‘criticism’. So this dichotomy can be rephrased as: ‘has pending criticisms’ vs ‘has no pending criticisms’, or ‘has reasons to be rejected’ vs ‘has no reasons to be rejected’. Note that there’s a difference: if you think some idea is bad, you submit a criticism. If you think it’s good, you can still submit a criticism because it might not yet be as good as you want it to be. So regardless of how good a theory might be, it can still have pending criticisms, and thus reasons to reject it. Think of Newtonian physics, which (I’m told) is a superb theory, but it’s false and (as I understand it) has plenty of pending criticisms.
‘Has pending criticisms’ vs ‘has no pending criticisms’ is directly comparable whereas ‘good’ and ‘bad’ aren’t directly comparable. And ‘has n pending criticisms’ vs ‘has m’ or ‘has 0 pending criticisms’ are even numerically comparable.
Veritula does not implement Deutsch’s epistemology. It implements Popper’s. I don’t think they’re compatible.
(As an aside, I’m not sure how I could implement Deutsch’s epistemology even if I wanted to. Would I give each idea a slider where people can say how ‘good’ the idea is? What values would I give the slider? Would the worst value be -1,000 and the best +1,000? How would users know to assign 500 vs 550? Would a ‘weak’ criticism get a score of 500 and a ‘strong’ one 1,000? What if tomorrow somebody finds an even ‘stronger’ one, does that mean I’d need to extend the slider beyond 1,000? Do I include arbitrary decimal/real numbers? Is an idea’s score reduced by the sum of its criticisms’ scores? If an idea has score 0, what does that mean – undecided? If it has -500, does that mean I should reject it ‘more strongly’ than if it had only -100? And so on. Deutsch says you haven’t understood something if you can’t program it, and I don’t think he could program his epistemology.)
Great clarification of Popper’s position—and of how it differs from Deutsch’s. Very insightful. I see what you mean about there being no room for positive arguments, and that labeling explanations as good or bad can itself be a form of positive argument. Still, I find value in the distinctions Deutsch makes when describing theories as good, bad, better, fundamental, deep, or anti-rational. Unfortunately, I don’t yet know how to reconcile that, nor do I have a satisfactory alternative theory or criticism to offer. I’d like to revisit this later, but I have a busy stretch coming up, and I expect the discussion will take some real research and time—so I’d rather not start it just yet.
For now, I’ll leave a few breadcrumbs (mostly notes to self) to pick up later.
Deutsch’s idea of a “good explanation” seems to involve the following elements:
Structure: Whether the explanation is built the right way—that is, whether it describes the mechanism of how and why something works rather than appealing to authority, source, or mere results. The “hard-to-vary” criterion also seems tied to this structural quality.
Resilience: Whether it stands up to repeated criticism and testing. Deutsch agrees that general relativity and quantum theory are “wrong” in the Popperian sense—since they have gaps or domains where they fail—yet he still counts them among our best explanations because they’ve been repeatedly tested and shown to work reliably within their applicable domains.
Depth and reach: A hallmark of a good explanation is that understanding it allows you to understand a range of other phenomena as well. Deutsch even suggests there’s a kind of convergence toward a unified theory of everything—hinting at a deep link between reality and its propensity to be explained.