Following http://lesswrong.com/lw/bwo/logical_fallacy_poster/ some people complained about

  • the sarcastic illustration
  • the lack of references
  • the weird categorization that should rather fit a Bayesian framework
  • the simplistic or even wrong definitions
  • and more

Yet this poster has ONE key difference with the ideal poster, it exists.

If it sparks criticisms that lead to a new, LessWrong compatible poster, then it is well worth the critics.


The obvious next step then is to make a poster that would allow to take into account such well founded suggestion and synthesize the LessWrong lessons visually.

In your opinion then what would be a good structure, e.g. a hierarchy of fallacies, and a design theme?

New to LessWrong?

New Comment


12 comments, sorted by Click to highlight new comments since:

Ahhh, from the title and vote count I expected this to be a presentation of said poster.

Still, great idea - having the GTD Workflow Map up in my office has (subjectively) improved my adherence to the model, I could see a rationality equivalent being very useful.

At least in my ideal poster-space this hierarchy would be included. Considering that knowing about biases can hurt people having a general theme of how to resolve arguments better rather than "here are some fallacies, avoid them/point them out" can't hurt.

You could make a top level distinction between problems with arguments (or how they are presented) and problems with thinking (or decision making), though there would be overlap. But I think, for a single poster, it might be overly ambitious to try to include both logical fallacies and cognitive biases.

And confusing - if someone makes use of a fallacy in an argument, is the bias in the person who is persuaded by the argument, in the person presenting the argument, or both?

From reading the other comments, this poster makes a three-way top level distinction.

Along similar-ish lines, it might be possible to use the hierarchy to score the quality of an argument. Essentially you'd assign a score from -4 to +4 for DH0 to DH7, then score an argument based on it's content. Although once you know it contains (say) a DH4 argument, you wouldn't keep on adding points for more DH4 arguments (otherwise an argument that was purely lots of DH4 statements would get a higher score than one DH7 statement).

It depends.

Usually you don't make use of a fallacy, it's more that you unknowingly commit one. If that's the case then the bias is in both you and the person who is persuaded by it.

On the other hand if you intentionally use a fallacy in order to persuade someone, you're a) dabbling in the dark arts and b) not actually biased (as you're not convinced by your own fallacy). However if you succeed in persuading someone with this method then that person would be the one with bias.

Is a poster of logical fallacies really something worth expending effort on? I'd find a crib-sheet of cognitive biases, illuminative cognitive disorders or helpful terminology way more useful. It'd need continuously updating, of course.

Much of the prep work would seem to be useful for the creation of both.

Yet this poster has ONE key difference with the ideal poster, it exists.

Have you seen the other poster?

http://www.obsidianfields.com/lj/venn-poster-large.jpg

There are also non-poster attempts at taxonomies: here and here

Perhaps one could adapt some of the work in what intelligence tests miss. For those that have the book: there is a chapter titled "How many ways can thinking go wrong? A Taxonomy of Irrational thinking tendencies and their relation to intelligence". That would give us the categories "heuristics", "lack of knowledge" and "wrong knowledge". The categories about knowledge could be further divided into academic fields such as probability theory, logic, language, psychology.

Is the aim of this to produce a poster of general fallacies that conform to LW standards, or to include fallacies specific to usual topics of LW discussion? E.g. anthropomorphizing evolution isn't something that would usually happen in any given debate, but happens often when discussing LW-related topics.

I was thinking about the first option "a poster of general fallacies that conform to LW standards" but I suppose the second could be useful too. Limited and probably more technical yet surely more useful here.

If such poster is created, it should be distributed to as many high schools as possible.

[-][anonymous]00

The first taxonomy that popped out at me:

Model doesn't match reality: false cause, slippery slope, gambler's fallacy,

Updating on weak evidence: similarity/analogy, familiarity/anecdote,

Argument on the basis evidence / justification / criteria that have been screened off or are irrelevant: genetic fallacy, appeal to authority, bandwagon, natural ideal, ad hominem, appeal to consequences

Equivocation: false dichotomous partition, ambiguity, strawman, external burden of proof (my priors are their priors?), circular argument

Restructuring theory without adjusting prior: special pleading, no true Scotsman, Texas sharpshooter

Premature compromise.

Also, Ulrike Hahn has a few papers explaining fallacies or trying to make a stronger normative theory of argumentation. Might be a good place to start if you want a more in depth chart.

[This comment is no longer endorsed by its author]Reply