Table of Contents
Kialo claims to be an online “debate platform powered by reason” and explains:
Kialo enables you to visualize discussions as an interactive tree of pro and con arguments. At the top is the thesis, which is supported or weakened by pro and con arguments underneath. Each one of these arguments can branch into subsequent arguments that support or attack them in turn.
I appreciate that Kialo said this clearly. It’s the mentality of indecisive arguments: partial arguments, degree arguments, credences or weighted factors. Critical Fallibilism (CF) claims that that viewpoint is believed by pretty much everyone but is incorrect. CF offers an alternative: decisive arguments and pass/fail evaluations. CF says to evaluate success or failure at specific goals rather than evaluating how good or bad ideas are on a spectrum.
The standard view also incorrectly evaluates ideas for broad, vague goal like “get money” instead of specific goals like “get $30” or “get $50,000” . One of the main reasons for being unable to reach decisive conclusions is that they’re essentially trying to reach conclusions about multiple different goals at the same time. “Get money” is really a group of goals to get different amounts of money, and a particular outcome (like getting $500) succeeds at some goals but fails at others, which looks like a mixed outcome (partial success) when you don’t distinguish between those different goals. Another goal people sometimes have is “get as much money as possible (within some constraints)”. Maximization goals are a single goal but don’t allow partial success because second place is not the maximum amount of money and is therefore, according to that goal, a failure. Maximization goals are usually a bad idea because there’s usually some “good enough” amount and anything above that is satisfactory.
Some people deny or fail to recognize how widespread weighted factor thinking is and that they personally do it. I often have to put effort into telling people what their own view is and explaining their own opinions to them, not just refuting their position and presenting an alternative. But most of those same people, who resist recognizing that CF disagrees with them, would read Kialo’s clear statement and see nothing wrong with it.
One of the issues is it’s convenient to talk about the supporting and weakening in terms of numeric point scores. I often talk this way to communicate more clearly. But people often don’t use numbers, so they fail to recognize it’s just a different way to talk about the same concepts. Kialo didn’t use numbers in their statement, but supporting or weakening ideas is equivalent to increasing or lowering their numeric score. It’s the same concept. And Kialo has “impact” scores based on people rating arguments from 0 to 4.
I’ll now give an explanation in terms of numeric points. Keep in mind it could all be translated into non-numeric words without changing the core conceptual meaning.
Kialo’s explanation reminded me of an important issue: The points-based view of arguments (which I think is justificationism) says that if you don’t answer a criticism (and maybe even if you do), then the criticism subtracts points from your side. Rather than answering a criticism, you can make a pro argument for why your side is good, and that can be equally or more effective. Often, you can gain more points for your conclusion by adding another pro to your side instead of giving a con that applies to their con. And it’s often easier to add more pros because that doesn’t require understanding or engaging with a critic’s perspective. And if you already made a bunch of strong pro arguments, so you have a ton of points already, then you can just ignore tons of criticism because you have spare points to lose. If you believe you have a significant lead in the debate, then you can just ignore arguments the other side gives, and all that happens (in your opinion) is that your lead gets less big. This stuff is bad and is part of why debates usually aren’t very productive.
This is part of why people don’t look at discussion trees like CF does. CF focuses attention on which nodes in the tree are unanswered, on which arguments are decisive or indecisive (for which goals), and on which factors are constraints and which aren’t. But with a system like Kialo, it’s normal for some idea to have several pros and several cons, and for some of those cons to be unanswered. That’s considered fine because the pros outweigh the unanswered cons. CF says a single decisive criticism is decisive, so every criticism needs to be answered – either with a direct rebuttal or by explaining that it’s indecisive.
Even if you do answer a criticism/con, it can still subtract some points. People expect counter-arguments to criticism to be partial and inconclusive, not decisive. So if a criticism subtracts 10 points from your side, you can make a positive argument for your side and gain 10 or maybe even 100 points. That’s a fully valid way to deal with criticism in the weighted factor view. And if you try to answer the criticism and you do a great job, you might counter 80% of it, so instead of doing 10 points of harm to your side it only does 2 point of harm to your side. So you only gained 8 points. That’s an inefficient way to get a good score.
In this system, there’s a general incentive to make higher level arguments over lower level arguments. A top level argument (a pro or con directly about the thesis) can be worth any number of points. But a con which is nested under some other node can only matter as much as its parent node. If an argument is adding 20 points, the very best a con for that argument could do is take those 20 points away. There is a ceiling on how much nested cons matter. Nested pros might have a similar issue: is their only purpose to counter cons they are siblings with? It’s actually unclear exactly how they work. Kialo, like most advocates of evaluating how strong arguments are, doesn’t try to give any specific or rigorous details of how to keep score in debates, with or without numbers.
In debate points systems, answering most criticism is a low value activity. It never lets you win debates. You should generally only try to answer a few high-scoring criticisms. It’s like how playing defense alone can’t win sports games. You need some offense, not just to limit the other team’s offense. CF, by contrast, says to look at which criticisms are decisive (would lead to failure if true), and to address them. CF says we learn by error correction not by building up positive knowledge with error-avoiding methods nor by ignoring problems to focus on positive claims.
Mixing Decisive and Indecisive Arguments
The people who believe in argument strengths and indecisive arguments usually also accept the existence of decisive arguments which conclusively refute things (and sometimes they also believe in decisive arguments that prove things, which is often infallibilist). But they usually view those arguments as an uncommon case that doesn’t merit much thought. One of people’s favorite decisive arguments is “If it contradicts the evidence, it’s wrong.” (Richard Feynman said something like that.) But there are many ideas that don’t contradict the evidence, so further analysis is needed. The general thinking is that once you get rid of the decisively refuted ideas, then you have a lot of reasonable ideas left (that disagree with each other) which you have to analyze with indecisive arguments. So looking at argument strengths is a tiebreaker for the ideas which aren’t decisively refuted.
If indecisive arguments are a secondary tiebreaker, that viewpoint actually acknowledges a lot about decisive arguments. It means decisive arguments exist, they logically work, we have some, they’re better, and they have priority. Indecisive arguments are an inferior fallback option for when we run out of decisive arguments. If people would think that through, then next they should consider what the limits on decisive arguments are. Instead of just accepting that we don’t have enough of them, why not look for more? Are there some fundamental limits that make it impossible to get plenty of decisive arguments? If so, what are those limits? There’s something to study and understand there. Or maybe it’s possible but for some reason it’s so difficult that it’s impractical. Or maybe it appears difficult now, but if people actually tried to do it and practiced it, they’d get better at it and start finding it pretty easy. But people have shown little interest in this topic and haven’t investigated much. That seems to be because it hasn’t really been thought through and clearly laid out.
Suppose settling all debates with decisive arguments is theoretically possible, but too hard and impractical. Then we could think of points-based debate systems as an approximation or shortcut for an unreachable ideal. If that’s what they are, that’d be worth knowing.
If we could find more decisive arguments, then it’d be reasonable to consider using them in a bigger role in debates. If we could find enough, maybe we should use them exclusively. That’d be worth considering. And that is actually CF’s position: we can find enough decisive arguments to use them exclusively, and we should do that (also, indecisive arguments are worse than typically believed). CF says the way to begin finding more decisive arguments is by using specific, unambiguous goals (and often considering multiple goals, and giving ideas different evaluations for different goals, instead of just considering one goal and trying to give each idea a single evaluation).
There are some difficulties with a debate tree with decisive arguments in it. Basically, a decisive argument wins the debate if ignored, so you have to answer it. (If it’s defending against a criticism, then a decisive argument may only win a branch of the discussion, a subtree, and invalidate that criticism.) But if you have to answer all the arguments, that’s a lot of work. And how will a debate end?
If you answer arguments with some decisive counter-points, now the other guy is forced to answer your new arguments or he loses. When he does that, now you have to give more arguments. Every time someone adds one new decisive argument to the debate, the other side has a burden to answer it or lose. This could create endless work. Basically people will keep having to answer arguments until someone gives up and concedes.
Suppose two people debate and they each say one argument at a time. And they always respond to the other person’s latest point. Then the debate is structured like a linear chain, not a branching tree. In this case, if they only use decisive arguments, then whoever gets the last word wins. The conclusion of the debate is unstable. When one guy speaks, then the conclusion favors him. Then the other guy speaks and the conclusion favors him. And it keeps going back and forth until someone stops.
What can be done about these difficulties?
Well, as we discussed above, decisive arguments can be hard to come by. A debate shouldn’t go back and forth indefinitely with decisive arguments because people will run out of them. As long as people have important good points to say, it’s good for a debate to continue. But then, before too long, someone should run out of arguments, admit it, and concede the debate.
Conceding doesn’t have to mean agreeing with the other guy and adopting his view. You could concede by saying “I don’t know of a decisive refutation of that argument, so it looks like you might be right. I’ll have to think it over more, and in the meantime I’ll stop acting on my theory, since I acknowledge it might be wrong.” When you concede a debate, you should usually become uncertain rather than immediately trying to believe what your opponent believes. Perhaps you were both wrong. And even if he was right, you’ll need to study it some in order to understand it better before it starts being your own opinion. And if you think it over more, you may come up with a new argument which saves your original conclusion.
Overall, debates using decisive arguments should reach an endpoint reasonably quickly because people have some decisive arguments but not an unlimited number of them. CF claims we can come up with enough decisive arguments to reach conclusions rather than being stuck (or trying to resort to somehow reaching conclusions using indecisive arguments). That doesn’t mean endless debates.
People may argue in bad faith and come up with an unlimited number of dumb arguments that claim to be decisive arguments. They may also find some way of generating many arguments which isn’t bad faith but isn’t helpful. What can you do about that? Make an argument which applies to a whole category or pattern of ideas. When your argument applies to more than one thing, then it preemptively addresses some potential future arguments. You’ll be able to sometimes say “I already answered that” (and give a link, quote or citation) instead of giving a new argument.
Using indecisive arguments doesn’t solve the problem that debates can take too long. It often makes it worse because it encourages people to come up with as many arguments for their side as possible, even pretty bad arguments. It rewards argument quantity because even a poor argument is worth a few points.
To make debates shorter, we need to focus our attention on a small number of key issues. CF offers a way to do that: recognize that (in general) most factors have excess capacity and should be evaluated as “good enough” (in other words “pass” on a pass/fail scale). Only a few factors are near breakpoints and are therefore key issues that need more detailed attention.
Excluding all indecisive arguments also makes debates shorter because it reduces the number of arguments in debates. The reason indecisive negative arguments should be excluded is that they are (by definition) compatible with the thing they’re arguing against actually being correct. A criticism should contradict the thing it’s criticizing; if it’s an indecisive criticism – if it’s compatible with its target being true – then in what sense is it even a criticism? What does it matter? Indecisive positive arguments are reasons some idea is good which are compatible with it being bad, so again they don’t really matter.
People often have an intuition that some argument matters but they don’t know how to make the argument decisive. What they should do is start stating goals it is decisive for. As a logical matter, all arguments are decisive for some goals. So brainstorm what goals your point is actually decisive for and then consider if those goals matter. Often, once you state a goal that it matters too, you’ll immediately recognize that you do or don’t care about that goal. Keep in mind that the stated goals in a discussion may not correspond well to people’s actual goals; it’s common to improve our explicit understanding of our goals.
Contradictions on the Same Side
Another problem with Kialo is that no individual takes responsibility for a particular side or case. The result is that many arguments on the same side contradict each other. Even if figuring out how strong the arguments on each side are was a valid method, you couldn’t do it like this. You’d need to consider multiple separate cases for and against the thesis. Each case would have internal consistency. Then you’d take each pro case and pair it with each con case and figure out which case is stronger. You’d end up with multiple conclusions, not a single conclusion. If you’re lucky, one pro or con case will beat every opposing case, in which case you can reach a single conclusion. However, sometimes that won’t happen. Pro case p1 beats con case c1 but loses to con case c2. However, con case c2 loses to p2, p3 and p4, but also of those lose to c1.
Kialo gives an example argument defending free will by claiming that thought and consciousness are immaterial and therefore are not subject to the laws of physics. A different view is that brains are literal computers which follow the laws of physics, and free will, thought and consciousness are all emergent properties. That view reaches the same conclusion (free will exists) in a contradictory way.
If you take any major political controversy, you can find people reaching the same conclusion for different, incompatible reasons. Every presidential candidate has voters who like them for a variety of different reasons, some of which contradict each other. Some people voted for Trump because they thought he was an alt-right white supremacist, while others voted for Trump because they thought he was a moderate, 1990’s center-leftist who hadn’t moved further left since then. You shouldn’t add together those two contradictory claims about why Trump should be president; they don’t work together to strengthen his case.
Some people want to lower minimum wage because they think poor people don’t deserve that much money, while others want to lower minimum wage in order to help create more jobs to help low socioeconomic status people. It doesn’t make sense to add together those viewpoints and say that they both strengthen the case for lowering minimum wage. They should be seen as separate, incompatible cases, not part of the same side/team/case.
So even if you could add up partial arguments for one side, you couldn’t just let multiple authors add contradictory arguments for the same side and add all those up. Sometimes a single author makes a contradictory case, but it’s a much more common problem when there are multiple people providing arguments.
A further problem with Kialo’s multi-author approach is they end up with big, cluttered debates and then delete some of what people say. But there’s no way to prune a discussion, and delete some arguments for a side, which satisfies all advocates of that side. Some people on that side get left out, have some of their points removed, and lack representation. Not everyone can be represented in a single tree that’s meant to be readable in full, and the system is set up to instead represent a mix of the more popular views, even if they contradict, rather than to explore all the ideas. It’d be better if they handled multiple separate cases for or against a thesis, instead of trying to merge all pro or con cases into one and then deleting a lot of stuff because there’s way too much.
Kialo actually has a further problem where specific, individual arguments are written by groups of people who negotiate (like design by committee), rather than an individual. Kialo also appears to want arguments to be short, but sometimes important arguments are long and detailed (and should actually be a whole subtree instead of a single node).